AMD's Future Chips & SoC's: News, Info & Rumours.

[8350rocks]

I think we missed another geekbech leak they link in here: http://techreport.com/news/32558/rumor-amd-raven-ridge-apu-scores-appear-on-geekbench

This is the other RR leak: https://gfxbench.com/device.jsp?benchmark=gfx40&os=Windows&api=gl&D=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&testgroup=overall

I think the closest one from Intel would be...

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&D2=Intel%28R%29+Core%28TM%29+i7-7660U+CPU+with+Iris%28TM%29+Plus+Graphics+640

It's even an Iris part. I'm actually liking this information. I just hope AMD doesn't go nuts with pricing >_<

This info was posted before RyZen 7U vs i5U

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Radeon%28TM%29+Vega+10+Mobile+Graphics&D2=Intel%28R%29+Core%28TM%29+i5-7260U+CPU+with+Iris%28TM%29+Plus+Graphics+640

Any reason why the i5 has way better scores than the i7? 😵

There's a huge discrepancy between the two.

i7 could be thermal throttling...or the i5 could be on an older version of geekbench that favored Intel extensions more heavily.

 

[-Fran-]

I think we missed another geekbech leak they link in here: http://techreport.com/news/32558/rumor-amd-raven-ridge-apu-scores-appear-on-geekbench

This is the other RR leak: https://gfxbench.com/device.jsp?benchmark=gfx40&os=Windows&api=gl&D=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&testgroup=overall

I think the closest one from Intel would be...

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&D2=Intel%28R%29+Core%28TM%29+i7-7660U+CPU+with+Iris%28TM%29+Plus+Graphics+640

It's even an Iris part. I'm actually liking this information. I just hope AMD doesn't go nuts with pricing >_<

This info was posted before RyZen 7U vs i5U

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Radeon%28TM%29+Vega+10+Mobile+Graphics&D2=Intel%28R%29+Core%28TM%29+i5-7260U+CPU+with+Iris%28TM%29+Plus+Graphics+640

Any reason why the i5 has way better scores than the i7? 😵

There's a huge discrepancy between the two.

i7 could be thermal throttling...or the i5 could be on an older version of geekbench that favored Intel extensions more heavily.

Looking at the information of each, the i7 does seem to have "extensions" in it... Whatever that means. Other than that, I could not spot differences... The i7 has higher turbo even. That score is weird, to say the least.

Cheers!

 

[goldstone77]

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

 

[adamsleath]

I think we missed another geekbech leak they link in here: http://techreport.com/news/32558/rumor-amd-raven-ridge-apu-scores-appear-on-geekbench

This is the other RR leak: https://gfxbench.com/device.jsp?benchmark=gfx40&os=Windows&api=gl&D=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&testgroup=overall

I think the closest one from Intel would be...

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&D2=Intel%28R%29+Core%28TM%29+i7-7660U+CPU+with+Iris%28TM%29+Plus+Graphics+640

It's even an Iris part. I'm actually liking this information. I just hope AMD doesn't go nuts with pricing >_<

Cheers!

It's definitely a huge improvement on previous generation. It's always about the price once a device does what you need it too!

Geekbench benchmarks indicate a 37% increase on single-core testing with a Ryzen 5 2500U APU compared with AMD’s top 7th generation APU, the A12-9800. Raven Ridge also performs 49% better in multi-core testing against the last generation, which is expected to have performance gains thanks to Raven Ridge’s extra threads.

AMD’s last generation A12-9800 is a desktop chip, and capable of running at up to 4.2GHz, while the Ryzen 5 2500U is supposedly easily defeating it’s 65W TDP last-gen rival at only 2.0GHz. For a mobile APU, the 2500U should also feature a much lower TDP than its desktop counterpart.

Performance benchmarks show AMD's Raven Ridge steamrolling 7th Gen APUs

Edit: On a side note: I've remember years ago talking to some friends about ARM and Android. They were talking about wait till Intel gets cpu's in there. I said, price will always dictate what people will buy based on needs and wants. Needs will win out if the wants don't matter. Basically, as long as the screen and apps are responsive to fit you needs, performance beyond that is wasted and unnecessary. Not having better benchmarks doesn't necessarily mean that a product isn't good enough to do the job incredibly well.

totally agree with that. if the price is right. sells like hotcakes. never underestimate the price savvy consumer. with affordable multicore and performance so close the gap is irrelevant. = amd success. competition and product diversity, gotta love it.

 

[juanrga]

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

 

[goldstone77]

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

 

[juanrga]

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

Just as I said, Glofo 7nm is not that magic node that certain analyst pretended (I will not gave names, cooff scoutgh cooff!!! )... And AMD was forced to pay twice (fixed penalty plus variable penalty) for having rights to use other foundries tech. Evidently AMD wouldn't do that if Glofo 7nm was the second coming of...

Despite all the hate I received for saying facts, reality proves I was right... once again.

 

[goldstone77]

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

Just as I said, Glofo 7nm is not that magic node that certain analyst pretended (I will not gave names, cooff scoutgh cooff!!! )... And AMD was forced to pay twice (fixed penalty plus variable penalty) for having rights to use other foundries tech. Evidently AMD wouldn't do that if Glofo 7nm was the second coming of...

Despite all the hate I received for saying facts, reality proves I was right... once again.

This doesn't mean that GlobalFoundries 7nm is going to be bad. If you have been keeping up with Vega Mega thread there is huge problem with supply, which is no doubt one of the reasons for the switch. It's is discussed by Gamers Nexus.

Edit: Supply side was bad they had 2 different sources, and issues with HBM2 packaging.

 

[goldstone77]

AMD Ryzen 5 2500U APU With Vega Graphics Spotted in Geekbench Benchmarks
September 16, 2017 - 03:32 AM | Tim Verry

Back in May AMD made Ryzen Mobile official indicating that the APUs previously known as "Raven Ridge" would be launching in the second half of 2017. As that launch window closes, more details are starting to trickle out including benchmarks scores. The latest appearance of Raven Ridge is in a Geekbench benchmark score results page where a "Ryzen 5 2500U" APU achieves a single core score of 3,561 and a multi-core score of 9,421. These are fairly impressive results on their own, but especially considering that Ryzen Mobile chips are reportedly using up to 50% less power versus last generation Bristol Ridge processors while handily beating them in performance offered.

AMD has previously claimed that its Ryzen Mobile (Raven Ridge) APUs will offer up to 50% more CPU performance and 40% more GPU performance compared to its 7th Generation APUs. The leaked Geekbench scores seem to back up those claims (for the most part) with the Ryzen 5 2500U scoring slightly above 36% better single core performance and 48% better multi-core performance compared to the AMD A12-9800 APU with the latter being due primarily to the addition of SMT to the processor design allowing for twice the number of CPU threads (eight total). The performance improvements are also due to the move from Excavator to a Zen-based design on a smaller more power efficient process node. What is most impressive about this mobile part is that it is that much faster than a 65W quad core (4 core / 4 thread) desktop Bristol Ridge APU clocked at 3.8 GHz base and 4.2 GHz boost while using approximately half the power!

The Geekbench benchmark is only one data point, but is still a positive sign. One thing it does not reveal is clockspeed as while it lists 2.0 GHz that number is likely only the base and not the maximum boost frequency. Further, details on the Vega-based GPU are still unknown although the Infinity Fabric should help tremendously in reducing the bottleneck and keeping the on die GPU fed with data while gaming especially when paired with fast dual channel memory or HBM (I just hope that Ryzen Mobile is not held back like previous generation mobile APUs were with laptop manufacturers pairing them with single channel memory setups). We also do not know officially the number of stream processors that will be included in any of the Vega GPUs used in Ryzen Mobile with past rumors going up to 1024 SPs (mobile parts will likely be capped at 512 or 768 at the extreme). AMD claims that Ryzen Mobile will offer up to 40% more GPU performance, which to me suggests that we will possibly see higher GPU core counts but for the most part performance improvements are going to come from architecture improvements.

I am really interested to see how Raven Ridge plays out and hope that it is one step closer to finally realizing that HSA future AMD has been promsing me for years!

 

[goldstone77]

Foundry Roadmaps: Real Solutions, Or Just Hedging?
JUNE 19TH, 2017 - BY: JOANNE ITOW

In May, Samsung revealed a full menu of processes that have been or will be released over the next few years. The new processes will be rolled out each year to provide new and existing customers with a variety of options to meet the competitive and broad product market application needs. In addition to their 10nm processes already in production, Samsung plans to offer risk production of 8LPP this year, 7LPP in 2018, 6LPP and 5LPP in 2019 and 4LPP in 2020. Samsung also will continue to offer a fully depleted SOI process. In fact, their roadmap includes an 18FDS, to be released in 2019.

Prior to Samsung’s Foundry Forum, TSMC provided a review of their process options at the TSMC Technology Symposium. In addition to 10nm being ramped this year, TSMC also has numerous new process technologies going back to 22nm and 12nm, as well as specialty processes for automotive, MEMS and CMOS image sensors.
At the April 2017 Technology Forum, TSMC announced they had 12 tapeouts lined up for 7nm in 2017. They also announced 8 design wins for the 16FFC automotive platform.

Most recently, GlobalFoundries announced the availability of its 7nm (7LP) FinFET semiconductor technology. Design kits are available now, and the first customer products based on 7LP are expected to launch in the first half of 2018, with volume production ramping in the second half of 2018.

All three major foundries—TSMC, GF and Samsung—have publically addressed their rollout of EUV. In March 2017, TSMC provided their update on EUV showing successful, continuous throughput of 1400 wafers/day/machine at 125W source. They are confident that they will achieve acceptable throughput at 250W for their 7nm+ process. N7+ will include EUV layers and will be ready by 2018.

Samsung adopting the EUV before everyone looks like it's going to pay off for them having 6,5LPP in 2019, and 4LPP in 2020! Amazing things are just around the corner.

 

[aldaia]

I think we missed another geekbech leak they link in here: http://techreport.com/news/32558/rumor-amd-raven-ridge-apu-scores-appear-on-geekbench

This is the other RR leak: https://gfxbench.com/device.jsp?benchmark=gfx40&os=Windows&api=gl&D=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&testgroup=overall

I think the closest one from Intel would be...

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&D2=Intel%28R%29+Core%28TM%29+i7-7660U+CPU+with+Iris%28TM%29+Plus+Graphics+640

It's even an Iris part. I'm actually liking this information. I just hope AMD doesn't go nuts with pricing >_<

This info was posted before RyZen 7U vs i5U

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Radeon%28TM%29+Vega+10+Mobile+Graphics&D2=Intel%28R%29+Core%28TM%29+i5-7260U+CPU+with+Iris%28TM%29+Plus+Graphics+640

Any reason why the i5 has way better scores than the i7? 😵

There's a huge discrepancy between the two.

i7 could be thermal throttling...or the i5 could be on an older version of geekbench that favored Intel extensions more heavily.

Looking at the information of each, the i7 does seem to have "extensions" in it... Whatever that means. Other than that, I could not spot differences... The i7 has higher turbo even. That score is weird, to say the least.

Cheers!

Those i5 scores are highly suspicious, considering that there are 2 entries for i5 7260U
https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=50817317&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=Intel%28R%29+Iris%28TM%29+Plus+Graphics+640&D2=Intel%28R%29+Core%28TM%29+i5-7260U+CPU+with+Iris%28R%29+Plus+Graphics+640

The second one has scores closer to i7
https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53460776&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=Intel%28R%29+Iris%28R%29+Plus+Graphics+640&D2=Intel%28R%29+Core%28TM%29+i7-7560U+CPU+with+Iris%28TM%29+Plus+Graphics+640

 

[juanrga]

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

Just as I said, Glofo 7nm is not that magic node that certain analyst pretended (I will not gave names, cooff scoutgh cooff!!! )... And AMD was forced to pay twice (fixed penalty plus variable penalty) for having rights to use other foundries tech. Evidently AMD wouldn't do that if Glofo 7nm was the second coming of...

Despite all the hate I received for saying facts, reality proves I was right... once again.

This doesn't mean that GlobalFoundries 7nm is going to be bad. If you have been keeping up with Vega Mega thread there is huge problem with supply, which is no doubt one of the reasons for the switch. It's is discussed by Gamers Nexus.

Edit: Supply side was bad they had 2 different sources, and issues with HBM2 packaging.

Vega supply problems don't have anything to do with Glofo. So that is not the reason for the switch to TSMC. I already said why AMD was going to switch to TSMC for 7nm, and I did before it happened.

 

[goldstone77]

Spoiler

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

Just as I said, Glofo 7nm is not that magic node that certain analyst pretended (I will not gave names, cooff scoutgh cooff!!! )... And AMD was forced to pay twice (fixed penalty plus variable penalty) for having rights to use other foundries tech. Evidently AMD wouldn't do that if Glofo 7nm was the second coming of...

Despite all the hate I received for saying facts, reality proves I was right... once again.

This doesn't mean that GlobalFoundries 7nm is going to be bad. If you have been keeping up with Vega Mega thread there is huge problem with supply, which is no doubt one of the reasons for the switch. It's is discussed by Gamers Nexus.

Edit: Supply side was bad they had 2 different sources, and issues with HBM2 packaging.

Vega supply problems don't have anything to do with Glofo. So that is not the reason for the switch to TSMC. I already said why AMD was going to switch to TSMC for 7nm, and I did before it happened.

Vega supply problems don't have anything to do with Glofo.

Source?

So that is not the reason for the switch to TSMC. I already said why AMD was going to switch to TSMC for 7nm, and I did before it happened.

Source?

 

[juanrga]

Spoiler

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

Just as I said, Glofo 7nm is not that magic node that certain analyst pretended (I will not gave names, cooff scoutgh cooff!!! )... And AMD was forced to pay twice (fixed penalty plus variable penalty) for having rights to use other foundries tech. Evidently AMD wouldn't do that if Glofo 7nm was the second coming of...

Despite all the hate I received for saying facts, reality proves I was right... once again.

This doesn't mean that GlobalFoundries 7nm is going to be bad. If you have been keeping up with Vega Mega thread there is huge problem with supply, which is no doubt one of the reasons for the switch. It's is discussed by Gamers Nexus.

Edit: Supply side was bad they had 2 different sources, and issues with HBM2 packaging.

Vega supply problems don't have anything to do with Glofo. So that is not the reason for the switch to TSMC. I already said why AMD was going to switch to TSMC for 7nm, and I did before it happened.

Vega supply problems don't have anything to do with Glofo.

Source?

So that is not the reason for the switch to TSMC. I already said why AMD was going to switch to TSMC for 7nm, and I did before it happened.

Source?

Digitimes.

Myself.

 

[gamerk316]

Spoiler

Coffee Lake Coming With 1151 Socket, But Still Requires 300-Series Chipset
by Paul Alcorn September 8, 2017 at 1:05 PM

AMD plans to support its Socket AM4 for all of its processors until 2020, so Intel’s lack of backward compatibility with existing 200-series chipsets will likely dominate the conversation.

How about that for an upgrade path!

That AM4 upgrade path is not confirmed. The only confirmed AM4 socket processor is Pinnacle Ridge. And even if it is finally confirmed that AM4 will last until 2020, AM4 customers will miss updates as DDR5 and PCIe4.

Scroll up

Intel and AMD are the key enablers for the broad desktop market; we certainly won't see PCIe 4.0 GPUs and SSDs without a slot to plug them into. AMD has slated PCIe 4.0 for 2020. We imagine Intel is also chomping at the bit to deploy PCIe 4.0 3D XPoint and NVMe SSDs, but the company remains silent on its timeline.

Yeah it's still far to early for DDR5 and PCI-E 4.0 to become mainstream anyways. They can't even get DDR4 ram prices in check last thing they should be worrying about is a newer standard.

AM4 will probably be around until 2020 and with Zen3. PCI-E 4.0 isn't even needed with a 1080Ti.

Technically, PCI-E 1.0 x16 is enough for a single 1080Ti. Barely, but still, it's not like we're hurting for bandwidth over the bus. The main benefits for newer PCI-E specs are more for lower latency, which have larger benefits for hardware other then GPUs (which aren't really that latency sensitive all things considered; 16ms is an eternity in computing).

I believe PCI-E 4.0 they are also pushing the removal of PCI-E power connectors, and have the socket supply all the power. And of course NVME.

I think we missed another geekbech leak they link in here: http://techreport.com/news/32558/rumor-amd-raven-ridge-apu-scores-appear-on-geekbench

This is the other RR leak: https://gfxbench.com/device.jsp?benchmark=gfx40&os=Windows&api=gl&D=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&testgroup=overall

I think the closest one from Intel would be...

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Ryzen+7+2700U+with+Radeon+Vega+Graphics&D2=Intel%28R%29+Core%28TM%29+i7-7660U+CPU+with+Iris%28TM%29+Plus+Graphics+640

It's even an Iris part. I'm actually liking this information. I just hope AMD doesn't go nuts with pricing >_<

This info was posted before RyZen 7U vs i5U

https://gfxbench.com/compare.jsp?benchmark=gfx40&did1=53156431&os1=Windows&api1=gl&hwtype1=iGPU&hwname1=AMD+Radeon%28TM%29+Vega+10+Mobile+Graphics&D2=Intel%28R%29+Core%28TM%29+i5-7260U+CPU+with+Iris%28TM%29+Plus+Graphics+640

Any reason why the i5 has way better scores than the i7? 😵

There's a huge discrepancy between the two.

Thermals and suboptimal threading come to mind.

 

[aldaia]

Interesting AMD EPYC 7601 review on Ubuntu Linux
http://www.phoronix.com/scan.php?page=article&item=epyc-7601-linux&num=1

$4200 32 core single socket vs $5200 40 core dual socket

Since 7601 was tested on single socket, actually almost same performance can be expected from the much cheaper $2100 7551P

 

[juanrga]

Interesting AMD EPYC 7601 review on Ubuntu Linux
http://www.phoronix.com/scan.php?page=article&item=epyc-7601-linux&num=1

$4200 32 core single socket vs $5200 40 core dual socket

Since 7601 was tested on single socket, actually almost same performance can be expected from the much cheaper $2100 7551P

The 7551P has lower performance than the 7501. The Intel config is also a bit weird, the chip selected is a quad-processor.

There are redflags on all the review. For instance:

The dual Xeon Gold 6138 system understandably comes out ahead given it has 80 threads versus 64 on the EPYC 7601. While this dual Xeon Gold system has 1.25x the thread count of the EPYC system, its EP.C performance is only better by 1.12x.

He forgot to mention in that phrase that the Intel chip has lower clocks. In fact if we divide the tread count by the clock deficit we can explain the performance gap.

 

[8350rocks]

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

Just as I said, Glofo 7nm is not that magic node that certain analyst pretended (I will not gave names, cooff scoutgh cooff!!! )... And AMD was forced to pay twice (fixed penalty plus variable penalty) for having rights to use other foundries tech. Evidently AMD wouldn't do that if Glofo 7nm was the second coming of...

Despite all the hate I received for saying facts, reality proves I was right... once again.

The reason for the switch is GloFo is simply unable to match demand. AMD will not be buying fewer wafers than the agreement requires, hence no penalty.

 

[juanrga]

Spoiler

AMD To Change Suppliers for Vega 20 GPUs on 7nm, HBM2 Packaging for Vega 11
by Raevenlord Friday, September 8th 2017 15:15

AMD's RX Vega supply has seen exceedingly limited quantities available since launch. This has been due to a number of reasons, though the two foremost that have been reported are: increased demand from cryptocurrency miners, who are looking towards maximizing their single node hashrate density through Vega's promising mining capabilities; and yield issues with AMD's Vega 10 HBM2 packaging partner, Advanced Semiconductor Engineering (ASE). It's expected that chip yield for Vega 10 is also lower per se, due to it having a 484 mm² die, which is more prone to defects than a smaller one, thus reducing the amount of fully-enabled GPUs.

AMD's production partner, GlobalFoundries, has historically been at the center of considerations on AMD's yield problems. That GlobalFoundries is seemingly doing a good job with Ryzen may not be much to say: those chips have incredibly small die sizes (192 mm²) for their number of cores. It seems that Global Foundries only hits problems with increased die sizes and complexity (which is, unfortunately for AMD, where it matters most).
Due to these factors, it seems that AMD is looking to change manufacturers for both their chip yield issues, and packaging yield problems. ASE, which has seen a 10% revenue increase for the month of August (not coincidentally, the month that has seen AMD's RX Vega release) is reportedly being put in charge of a much smaller number of packaging orders, with Siliconware Precision Industries (SPIL), who has already taken on some Vega 10 packaging orders of its own, being the one to receive the bulk of Vega 11 orders. Vega 11 is expected to be the mainstream version of the Vega architecture, replacing Polaris' RX 500 series. Reports peg Vega 11 as also including HBM2 memory in their design instead of GDDR5 memory. Considering AMD's HBM memory history with both the original Fury and and now RX Vega, as well as the much increased cost of HBM2's implementation versus a more conventional GDDR memory subsystem, this editor reserves itself the right to be extremely skeptical that this is true. If it's indeed true, and Vega 11 indeed does introduce HBM2 memory to the mainstream GPU market, then... We'll talk when (if) we get there.

As to its die yield issues, AMD is reported to be changing their main supplier for their 7 nm AI-geared Vega 20 from GlobalFoundries to Taiwan Semiconductor Manufacturing Company (TSMC), who has already secured orders for AI chips from NVIDIA and Google. TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers. How this will affect AMD and GlobalFoundries' Wafer Agreement remains to be seen, but we expect AMD will be letting go of some additional payments GlobalFoundries' way.

Link to Gamers Nexus News talking about this.

Therefore my claims about Glofo 7nm and my prediction that AMD would switch to TSMC for GPUs is confirmed... 😀

TSMC's 7nm and CoWoS (chip-on-wafer-on-substrate) capabilities have apparently proven themselves enough for AMD to change manufacturers.

I said Lisa Su emphasized during investor day that they would use the leading edge technology. She wouldn't have renegotiated, and paid GlobalFoundries already if they didn't already have a plan for this. This is a win for AMD being able to go to the foundry with the best process technology.

Just as I said, Glofo 7nm is not that magic node that certain analyst pretended (I will not gave names, cooff scoutgh cooff!!! )... And AMD was forced to pay twice (fixed penalty plus variable penalty) for having rights to use other foundries tech. Evidently AMD wouldn't do that if Glofo 7nm was the second coming of...

Despite all the hate I received for saying facts, reality proves I was right... once again.

The reason for the switch is GloFo is simply unable to match demand. AMD will not be buying fewer wafers than the agreement requires, hence no penalty.

The reason for the switch is technical. And AMD already paid the fixed part of the penalty: about $300 million. It remains the variable part.

 

[jaymc]

Zen2 is 7nm.

TSMC 7nm and Glofo 7nm aren't compatible. AMD cannot use TSMC as "second source". I guess AMD will use TSMC for GPUs and Glofo for CPUs, because the Glofo process doesn't look suitable for GPUs.

They can use Samsung as a second source..
In fact it would be quite painless as the node is more or less a samsung node anyway.

 

[aldaia]

The 7551P has lower performance than the 7501.

NO

 

[juanrga]

Zen2 is 7nm.

TSMC 7nm and Glofo 7nm aren't compatible. AMD cannot use TSMC as "second source". I guess AMD will use TSMC for GPUs and Glofo for CPUs, because the Glofo process doesn't look suitable for GPUs.

They can use Samsung as a second source..
In fact it would be quite painless as the node is more or less a samsung node anyway.

Unlike what happens at 14nm where Glofo licensed the Samsung node and both foundries use 14LPP, the 7nm nodes are different and were developed for separate. Samsung 7nm is different to Glofo 7nm. Samsung uses EUV-L, Glofo uses multipatterning; Samsung has pitches of 54 and 36, Glofo has 56 and 40...

The nodes aren't compatible, AMD cannot use Samsung as second source. because would have to redo the design and make new masks...

AMD is using Glofo for CPUs and TSMC for GPUs.

 

[juanrga]

The 7551P has lower performance than the 7501.

NO

2.0/3.0GHz vs 2.2/3.2GHz plus binning.

 

[aldaia]

The 7551P has lower performance than the 7501.

NO

2.0/3.0GHz vs 2.2/3.2GHz plus binning.

Wrong

 

[jaymc]

Zen2 is 7nm.

TSMC 7nm and Glofo 7nm aren't compatible. AMD cannot use TSMC as "second source". I guess AMD will use TSMC for GPUs and Glofo for CPUs, because the Glofo process doesn't look suitable for GPUs.

They can use Samsung as a second source..
In fact it would be quite painless as the node is more or less a samsung node anyway.

Unlike what happens at 14nm where Glofo licensed the Samsung node and both foundries use 14LPP, the 7nm nodes are different and were developed for separate. Samsung 7nm is different to Glofo 7nm. Samsung uses EUV-L, Glofo uses multipatterning; Samsung has pitches of 54 and 36, Glofo has 56 and 40...

The nodes aren't compatible, AMD cannot use Samsung as second source. because would have to redo the design and make new masks...

AMD is using Glofo for CPUs and TSMC for GPUs.

Sorry I thought you were talking about Zen+ on 14nm+ my mistake