AMD CPU speculation... and expert conjecture

[8350rocks]

LOL!!!

 

[youcanDUit]

all i see are numbers and no real-life demos. how long do we have to wait?

 

[blackkstar]

AMD needs market share. With that the money will start flowing into their R&D (i hope) and the CEO yatches and beach houses, haha. Probably a big chunk will go to the later, lol.

Anyway, since AMD/ATI decided to stop going big with their dies, they've been able to properly compete against nVidia (not winning all the time, mind you), so I would not think the trend will change dramatically unless nVidia goes the same route as AMD. In fact, isn't GK104 precisely that?

Cheers!

GK114 is what you're referencing, IIRC, though that project was basically derailed from what I recall of semi accurate's discussion of upcoming NVidia projects.

In short, NVidia has no answer to Hawaii XT until Q4 2014, and it won't be something impressive that comes out by the time AMD refreshes again. In fact, if AMD plays their cards right, and launches new GPU architecture Q3 2014, then NVidia may release a product that isn't even competitive at that point when they do launch their new architecture the next quarter.

Wouldn't that be interesting? "Look! It's the new GTX 880 benchmarks! It's...it's...quite a bit slower than the R9-390X...well...it has shadowplay 2.0 and G-sync 2.0, and outdated HDMI ports though, so at least there's that...and it comes with a couple of games on Unreal Engine 3 that use PhysX."

Indeed. AMD beating a 551mm^2 chip with a 435mm^2 chip is proof that Nvidia engineers are not as good as AMD's.

Raja got hired in April 2013 and he hasn't even had enough time to do much to GCN yet. Hawaii is basically AMD's old guard doing very well.

If AMD can maintain this, it means that Nvidia will always be forced to sell larger, slower dies for more money.

I am just glad to see how AMD GPU division handles this. This could have ended up like FX 9590 where it's a "Titan Killer" for $999 and it would have fit in with Nvidia's PPD.

Hopefully it's a sign that AMD is working very hard on die size to performance ratios and that this kind of performance will trickle over.

Seeing a 4m/8c EX or SR CPU at 200mm^2 would be awesome.

 

[Ags1]

Spoiler

Spoiler

Since we're bringing up popularity on amazon, take a look here:
http://www.amazon.com/s/ref=sr_nr_n_0?rh=n%3A172282%2Cn%3A541966%2Cn%3A229189%2Ck%3Acpu&sort=popularity-rank&keywords=cpu&ie=UTF8&qid=1383430245&rnid=493964

FX 8350 most popular AMD CPU (these results include APUs, mind you).

We were talking about bestselling. The bestselling CPU on AMAZON is the FX-6300, followed by the FX-8350 both rank #87 and #88. There are Intel chips that rank higher. The higher I would find is the i5-4670K ranking #14

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_193870011_pg_5?_encoding=UTF8&pg=5

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_nav_e_2_541966

Filtering to only CPUs can give impression that the 8350 is selling well

http://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189/ref=zg_bs_nav_e_3_193870011

but overall it is 74 positions behing top-selling Intel.

Those are about components alone. If we count pre-build computers then total sales of APUS outperform CPUs by about 4:1. Last data I have FX-8000 series accounted for 2% of total desktop units shipped.

Do you see? AMD losing in laptops to Intel.

I already gave a link to how AMD lost laptop share to Intel. I add this

http://www.amazon.com/Best-Sellers-Electronics-Laptop-Computers/zgbs/electronics/565108/ref=zg_bs_nav_e_2_541966

The only way AMD can get a solid mobile win is if they can get into a Mac product. AMD already has wins to get into HEDT and Gaming PC markets.

Sure, there are rumors Apple is disappointed with Intel Haswell. Apple migrating to ARM is not possible now, therefore some people is speculating that AMD could get some Kaveri or Carrizo APU on future Apple laptop. Carrizo? Maybe. Kaveri? I doubt very much.

I have a lot of respect for Juaranga and I've known him for a while under different names in different forums, but I think he is wrong on this one. AMD needs to blow a lot of smoke to investors about going mobile and all that jazz. Analysts are screaming "OMG DESKTOPS ARE DEAD BUY 3 IPADS TO MULTITASK!!!" and AMD is in a position where they're doing very well in a subset of a dying market, when that subset is actually growing. It's like the Titanic is going down and AMD owns all the life boats.

But given the air around what analysts are saying, it'd be impossible to find investors if you go around saying that you're going to focus on PC Gaming. No one would want to invest in that. Not when desktop market share is shrinking overall (gaming PCs are doing better mind you but analysts seem to think that EVERYONE who owns a desktop can do everything they do on a desktop on a tablet) and new consoles are coming out.

There will be a dedicated, high end gaming platform for AMD that supports HSA over dGPU with a large CPU. AMD has been very careful in wording how they are scaling Mantle. They keep saying it scales to a large number of cores. If AMD were never going to release anything with more than 8 cores, they'd be calling it "scales up to 8 cores."

Well, I have a lot of respect for you and others in this thread. There are only two posters whom I lost the respect due to their posting style, but their names are easy to guess.

I don't think that any serious investor is saying that the desktop is dead, but are saying that desktops are a falling market. Those claims are based in sales data from multiple years plus predictions.

Apple is moving from desktops being their main revenue to laptops/tablets/phone being it now.

Intel and Nvidia are also moving to the tablet/phone market.

AMD was late to react to market changes, 'thanks' to myopic previous head.

Nobody says that desktops will disappear, but that desktops will remain as a niche market.

Parts of that niche market can grow. In fact the most optimistic predictions offer a 6.5% increase in PC gaming (!= whole desktop) for next year.

A 6.5% increase are peanuts for AMD. Why do you believe that AMD shares dropped by about a 15% one day after the Q3 results were made public? Because AMD still has an entire foot in the desktop (about a 50% of its finances still depend on the falling market) and the numbers don't match well still.

Sure you read AMD head claiming that they are moving from a past where most of its revenue did come from PCs to a future where most of its revenues don't come from PCs. AMD is now in a transition period with a 50%-50%.

Also nobody is saying that all desktops are being replaced by ipads. Desktops are being replaced by a mixture of laptops, tablets, TVboxes, hybrids and... desktops.

Finally, I note how some people here spend [strike]fantasizing[/strike] speculating about AM4, 10-core CPUs, triple dGPU configs... but avoids the talk given by AMD to OEMs a pair of days ago, specially the slide #13.

Here is the chart showing only CPUs: AMD has 4 of the top ten slots, ahead of most i3 and i5 CPUs:

http://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189/ref=zg_bs_nav_e_3_193870011

With i3s and i5s selling so poorly (worse even than FX!!!), I guess Intel will discontinue them 🙂

First, that link was given before by me in the same message that you are replying. Pay attention to this part of it:

Filtering to only CPUs can give impression that the 8350 is selling well

http://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189/ref=zg_bs_nav_e_3_193870011

but overall it is 74 positions behing top-selling Intel.

Indeed, as said above:

We were talking about bestselling. The bestselling CPU on AMAZON is the FX-6300, followed by the FX-8350 both rank #87 and #88. There are Intel chips that rank higher. The higher I would find is the i5-4670K ranking #14

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_193870011_pg_5?_encoding=UTF8&pg=5

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_nav_e_2_541966

Also, as said above:

Those are about components alone. If we count pre-build computers then total sales of APUS outperform CPUs by about 4:1. Last data I have FX-8000 series accounted for 2% of total desktop units shipped.

Moreover, I didn't say this in the message that you reply, but I said before in the thread: AMD is in a financial position very different than Intel. Intel can support making products for enthusiast and sell them at premium prices. AMD lacks the money for that. It only returned from the red one quarter ago.

People can continue [strike]fantasizing[/strike] speculating about a 10-core 5GHz SR FX CPU, but there is none.

I see no point in comparing CPU sales to thermal compound sales, financial calculator sales or RAM kits. If you include enough irrelevant products then you can keep the i5-4670 at 16 in the list and the 8350 will come in at 1153 in the list, right after organic fair trade potatoes.

The point is, if you really want to use Amazon data, then the FX chips handily outsell i3s and most i5s.

 

[sora]

point to note :

AMD is claiming a 30% gain (and i believe them) .
But a 30% gain over what ? Over Bulldozer/Piledriver.
So SR might just be competitive with Sandy's in single thread performance. And by then Haswell with 10% gain over IB will come to spoil the party. That is, if AMD can deliver SR in time......

I think they said that during bulldozer(Correct me if I'm wrong), which is actually a 32.5% gain.

Over what though? They aren't being very specific.

 

[juanrga]

The slide shows the APU giving ~92% of the FX-8350 performance in that gaming test. Say us which will be the performance in "multiplayer"?

athlon x4 750k = trinity. its found at the bottom by the pentium g's, overclocked to 4.5 ghz and still getting beat by 4350 @ 4.2 ghz and completely obliterated by the 8350.

you really need to quit trying to live by marketing slides. best case scenario != all cases.

While (Juanrga = wrong) {printf("JUANRGA IS NEVER WRONG")};

Let us forget that pclab is not neutral regarding AMD vs Intel, even accepting your non-marketing benchmark

trinity != richland

athlon != APU

GTX 780 != R9 280X

...

I would like to see multiplayer benchmarks in the same conditions that the benchmark in the slide #13. That is, I would like to see benchmarks of the new APU with the new APU drivers, the new APU optimized Windows, 2133 MHz ram, and paired with a R9 280X...

I estimate it would obtain ~70% of the performance of 8350.

 

[kansasboy001]

point to note :

AMD is claiming a 30% gain (and i believe them) .
But a 30% gain over what ? Over Bulldozer/Piledriver.
So SR might just be competitive with Sandy's in single thread performance. And by then Haswell with 10% gain over IB will come to spoil the party. That is, if AMD can deliver SR in time......

I think they said that during bulldozer(Correct me if I'm wrong), which is actually a 32.5% gain.

Over what though? They aren't being very specific.

That would make sense because amd's roadmaps say 15% increase per generation, but piledriver didn't really live up to the 15%. so a 30% increase over PD could be about right.

 

[juanrga]

Spoiler

look at slide 13 more closely, the title is MORE performance with MORE cores. AMD is still advertising that more performance = more cores. A 8 or 10 core steamroller is still very possible.

Thanks by being the first that comment on AMD last material.

Look at the benchmark. Approx.

A10-6970K: 53 FPS
FX-6350: 55 FPS
FX-8350: 57 FPS

Do you really believe that slide means that AMD will be releasing new 8 and 10-cores?

10-core makes no sense, therefore I am ignore this. New 8-core and 6-core CPU are still possible... as a Piledriver refresh a la Warsaw, but if that is the plan you dont give a talk that basically says: hey don't buy an 'expensive' FX-8350, because a cheap A10 makes almost the same work. You can obtain about a 92% of the gaming performance with half the money.

Please note that slide considers 1080p. At higher resolutions the differences will be small.

For me that slide solves the question made here about what will be AMD promoting for pairing with the new R9 Radeons. I said that AMD was migrating to APUs and several people here complained that AMD did need new CPUs for the new R9 Radeon.

Look at the slide. They are pairing the new R9 280X with the new A10-6970K APU. My bet is that during Kaveri presentation AMD will show benchmarks of the new Steamroller A10-7x00 APU with the R9 290X.

In a game that uses a GPU demanding engine...sure. Go ahead and benchmark Crysis 3 or BF4 multiplayer and see what it looks like.

8350 > APU

When the 8350 did suffer with older games you typical answer was benchmark newest games like crysis 3. Now your answer is benchmark older games like crysis 3... LOL

I am rather sure that AMD is relying heavily on BF4 lately, because it is the 'prototype' of next gen games. Next consoles having weak cores forces developers to offload heavy stuff to the GPU.

The slide shows the APU giving ~92% of the FX-8350 performance in that gaming test. Say us which will be the performance in "multiplayer"?

My last post was in response to your request for proof that SOI is that much better than bulk.

Since you clearly were unable to find the information on your own...(or didn't bother to look at every article comparing the 2 substrates)...I posted it for you, from multiple sources.

EDIT: This bodes well for GF:

http://semiaccurate.com/2013/10/29/ana-hunter-leaves-samsung-global-foundries/

There was no such request. My only request was to know if you are a GF PR or something because you look like one.

The rest of my post was a demonstration of how the 100MHz delta in OC between FX-6 and i5 destroy your 'theory' about how bad Intel chips OC due to choosing bulk.

I want to add that Kaveri ES clocked at 3.5GHz have been showed.

Crysis 3 and BF4 are not older games...are you feeling well?

Do you read? Because I have said you that BF4 is a new game. I have said more: I have said that it must be the prototype of the next gen games.

92% Performance in what? League of Legends? StarCraft? Diablo 2? Bioshock Infinite? MS Office? None of those is particularly demanding and falls directly in line with what I would expect. Let's talk about the examples where it's not 92%, and reality sets in and you realize it's really closer to 60% of the performance of the 8350, when the core counts matter.

Do you read? 92% of the performance in the BF4 benchmark given before.

Also, if you read the post I made before, the data was extremely relevant, Intel has had to tweak the process for the bulk wafers they use with tons of process advancements to the FinFET structures in order to compete in performance. Even then, when they try to aim for lower power, they end up sacrificing performance. All industry experts say FD-SOI will allow you to have both. So why not use it?

Contrary to your marketing, there is no expert agreement on that FD-SOI is better than bulk FinFET. In fact, the industry is moving towards FinFETs: Samsung, TMSC...

I am not GF PR, quite the opposite in fact. They have blundered things many times, however, they are AMD's last hope for a decent process for HEDT. So, like it or not, we have to hope that they get their act together to ensure that AMD does well in HEDT.

AMD returning ot SOI and Glofo is a secure way to suicide.

The simple fact of the matter is this...bulk provides no benefits, and many drawbacks. While FD-SOI provides many benefits, and virtually no drawbacks.

This explains why GloFO is #1 foundry in the world with hundred of customers waiting to use their SOI process production lines. whereas nobody uses bulk. GlofO 28nm SOI line is a complete success and is working at full production rate... no, wait, they have no customers still.

 

[juanrga]

Spoiler

Spoiler

Since we're bringing up popularity on amazon, take a look here:
http://www.amazon.com/s/ref=sr_nr_n_0?rh=n%3A172282%2Cn%3A541966%2Cn%3A229189%2Ck%3Acpu&sort=popularity-rank&keywords=cpu&ie=UTF8&qid=1383430245&rnid=493964

FX 8350 most popular AMD CPU (these results include APUs, mind you).

We were talking about bestselling. The bestselling CPU on AMAZON is the FX-6300, followed by the FX-8350 both rank #87 and #88. There are Intel chips that rank higher. The higher I would find is the i5-4670K ranking #14

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_193870011_pg_5?_encoding=UTF8&pg=5

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_nav_e_2_541966

Filtering to only CPUs can give impression that the 8350 is selling well

http://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189/ref=zg_bs_nav_e_3_193870011

but overall it is 74 positions behing top-selling Intel.

Those are about components alone. If we count pre-build computers then total sales of APUS outperform CPUs by about 4:1. Last data I have FX-8000 series accounted for 2% of total desktop units shipped.

Do you see? AMD losing in laptops to Intel.

I already gave a link to how AMD lost laptop share to Intel. I add this

http://www.amazon.com/Best-Sellers-Electronics-Laptop-Computers/zgbs/electronics/565108/ref=zg_bs_nav_e_2_541966

The only way AMD can get a solid mobile win is if they can get into a Mac product. AMD already has wins to get into HEDT and Gaming PC markets.

Sure, there are rumors Apple is disappointed with Intel Haswell. Apple migrating to ARM is not possible now, therefore some people is speculating that AMD could get some Kaveri or Carrizo APU on future Apple laptop. Carrizo? Maybe. Kaveri? I doubt very much.

I have a lot of respect for Juaranga and I've known him for a while under different names in different forums, but I think he is wrong on this one. AMD needs to blow a lot of smoke to investors about going mobile and all that jazz. Analysts are screaming "OMG DESKTOPS ARE DEAD BUY 3 IPADS TO MULTITASK!!!" and AMD is in a position where they're doing very well in a subset of a dying market, when that subset is actually growing. It's like the Titanic is going down and AMD owns all the life boats.

But given the air around what analysts are saying, it'd be impossible to find investors if you go around saying that you're going to focus on PC Gaming. No one would want to invest in that. Not when desktop market share is shrinking overall (gaming PCs are doing better mind you but analysts seem to think that EVERYONE who owns a desktop can do everything they do on a desktop on a tablet) and new consoles are coming out.

There will be a dedicated, high end gaming platform for AMD that supports HSA over dGPU with a large CPU. AMD has been very careful in wording how they are scaling Mantle. They keep saying it scales to a large number of cores. If AMD were never going to release anything with more than 8 cores, they'd be calling it "scales up to 8 cores."

Well, I have a lot of respect for you and others in this thread. There are only two posters whom I lost the respect due to their posting style, but their names are easy to guess.

I don't think that any serious investor is saying that the desktop is dead, but are saying that desktops are a falling market. Those claims are based in sales data from multiple years plus predictions.

Apple is moving from desktops being their main revenue to laptops/tablets/phone being it now.

Intel and Nvidia are also moving to the tablet/phone market.

AMD was late to react to market changes, 'thanks' to myopic previous head.

Nobody says that desktops will disappear, but that desktops will remain as a niche market.

Parts of that niche market can grow. In fact the most optimistic predictions offer a 6.5% increase in PC gaming (!= whole desktop) for next year.

A 6.5% increase are peanuts for AMD. Why do you believe that AMD shares dropped by about a 15% one day after the Q3 results were made public? Because AMD still has an entire foot in the desktop (about a 50% of its finances still depend on the falling market) and the numbers don't match well still.

Sure you read AMD head claiming that they are moving from a past where most of its revenue did come from PCs to a future where most of its revenues don't come from PCs. AMD is now in a transition period with a 50%-50%.

Also nobody is saying that all desktops are being replaced by ipads. Desktops are being replaced by a mixture of laptops, tablets, TVboxes, hybrids and... desktops.

Finally, I note how some people here spend [strike]fantasizing[/strike] speculating about AM4, 10-core CPUs, triple dGPU configs... but avoids the talk given by AMD to OEMs a pair of days ago, specially the slide #13.

Here is the chart showing only CPUs: AMD has 4 of the top ten slots, ahead of most i3 and i5 CPUs:

http://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189/ref=zg_bs_nav_e_3_193870011

With i3s and i5s selling so poorly (worse even than FX!!!), I guess Intel will discontinue them 🙂

First, that link was given before by me in the same message that you are replying. Pay attention to this part of it:

Filtering to only CPUs can give impression that the 8350 is selling well

http://www.amazon.com/Best-Sellers-Electronics-Computer-CPU-Processors/zgbs/electronics/229189/ref=zg_bs_nav_e_3_193870011

but overall it is 74 positions behing top-selling Intel.

Indeed, as said above:

We were talking about bestselling. The bestselling CPU on AMAZON is the FX-6300, followed by the FX-8350 both rank #87 and #88. There are Intel chips that rank higher. The higher I would find is the i5-4670K ranking #14

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_193870011_pg_5?_encoding=UTF8&pg=5

http://www.amazon.com/Best-Sellers-Electronics-Computer-Components/zgbs/electronics/193870011/ref=zg_bs_nav_e_2_541966

Also, as said above:

Those are about components alone. If we count pre-build computers then total sales of APUS outperform CPUs by about 4:1. Last data I have FX-8000 series accounted for 2% of total desktop units shipped.

Moreover, I didn't say this in the message that you reply, but I said before in the thread: AMD is in a financial position very different than Intel. Intel can support making products for enthusiast and sell them at premium prices. AMD lacks the money for that. It only returned from the red one quarter ago.

People can continue [strike]fantasizing[/strike] speculating about a 10-core 5GHz SR FX CPU, but there is none.

I see no point in comparing CPU sales to thermal compound sales, financial calculator sales or RAM kits. If you include enough irrelevant products then you can keep the i5-4670 at 16 in the list and the 8350 will come in at 1153 in the list, right after organic fair trade potatoes.

The point is, if you really want to use Amazon data, then the FX chips handily outsell i3s and most i5s.

We are comparing sales of CPUs to sales of CPUs. What happens is that if you filter by CPU alone you don't get an idea of the volume of sales for each. One processor is #1 and other is #4, but is the difference small or not? Did the #1 sell 10 units more or 10000 units more? The other link gives that: the top i5 is #14, the 8350 is #88. This gives an idea of the 8350 selling very bad.

Again, I emphasize those don't count pre-build computers.

In total numbers, the FX-8350 accounts for ~2% of the total of CPUs/APUs sold by AMD.

 

[juanrga]

point to note :

AMD is claiming a 30% gain (and i believe them) .
But a 30% gain over what ? Over Bulldozer/Piledriver.
So SR might just be competitive with Sandy's in single thread performance. And by then Haswell with 10% gain over IB will come to spoil the party. That is, if AMD can deliver SR in time......

I think they said that during bulldozer(Correct me if I'm wrong), which is actually a 32.5% gain.

Over what though? They aren't being very specific.

That would make sense because amd's roadmaps say 15% increase per generation, but piledriver didn't really live up to the 15%. so a 30% increase over PD could be about right.

The 15% claim for BD -> PD was net performance gain (IPC and clocks) and it was accurate

http://semiaccurate.com/2012/11/05/amds-bulldozer-core-compared-with-piledriver/

The 30% claim for BD --> SR is exclusively IPC.

The architectural change (and the performance boost) from PD to SR (~20% IPC) is much more important than from BD to PD (~8% IPC). The performance gain per generation is not lineal as this diagram clearly shows

I think the confusion comes from this other diagram, which is not representing performance, but performance per watt

Whereas SR will be ~20% faster (IPC) than PD, it will be only 10-15% more efficient than PD.

 

[UnrelatedTopic]

In a completely unrelated topic, this thread is a year old today so cheers

 

[-Fran-]

Let us forget that pclab is not neutral regarding AMD vs Intel, even accepting your non-marketing benchmark

trinity != richland

athlon != APU

GTX 780 != R9 280X

...

I would like to see multiplayer benchmarks in the same conditions that the benchmark in the slide #13. That is, I would like to see benchmarks of the new APU with the new APU drivers, the new APU optimized Windows, 2133 MHz ram, and paired with a R9 280X...

I estimate it would obtain ~70% of the performance of 8350.

Actually, Richland and Trinity are the same performance wise (same uArch). They didn't move anything or improved anything in regards to speed ups, just efficiency AFAIK.

And the Athlon II 750K IS a de-iGPU'ed APU. No L3 cache and the same speed with a little than you'd-like TDP because of the iGPU space. So comparing it with APUs is fair IMO. As long as the iGPU is not being used, of course.

Cheers!

 

[palladin9479]

8350 and juanrga just agree to disagree as both of you are just posturing with veiled insults being tossed in.

 

[de5_Roy]

Indeed. AMD beating a 551mm^2 chip with a 435mm^2 chip is proof that Nvidia engineers are not as good as AMD's.

? with hawaii, i see higher temps, power use per mm^2 compared to gk110, titan is not full gk110 afaik... gtx780 TI seems like the one. yields seem as bad as early titan (consumer gk110, not pro). i looked at newegg, no r9 290x available. amazon has 2, xfx and h.i.s. both are selling near and over $700. i wouldn't consider amd's 'victory' until supplies and prices stabilize. don't forget that amd is fighting with a whole new die and gpu against nvidia's silicon dump. seems like silicon dumping helps with yields and supplies.

titan's target market remains out of r9 290x's reach - [strike]really really high priced[/strike] premium smaller form factor gaming builds sold by boutique builders. may be... if r9 290 (non-x) can keep it's heat in control, time will tell.

Seeing a 4m/8c EX or SR CPU at 200mm^2 would be awesome.

i hope amd engineers can keep the cpu design small. zambezi and vishera are too big.

Spoiler

Actually, Richland and Trinity are the same performance wise (same uArch). They didn't move anything or improved anything in regards to speed ups, just efficiency AFAIK.

And the Athlon II 750K IS a de-iGPU'ed APU. No L3 cache and the same speed with a little than you'd-like TDP because of the iGPU space. So comparing it with APUs is fair IMO. As long as the iGPU is not being used, of course.

Cheers!

iirc trinity and richland are exactly same die size of 246 mm^2 (same die, possibly), similar electrical characteristics. richland is just better binned and firmware improved for better turbo and some other stuff. socket fm2 athlons are igpu-lasered-off apus.

New Steam Machine details revealed
http://techreport.com/news/25606/new-steam-machine-details-revealed

Valve is apparently "super interested" in Intel's Iris Pro IGP—and when asked if Intel or AMD could make processors with beefed-up integrated graphics specifically for Steam Machines, the firm said, "We don't have an answer that we can give you before January." That sure sounds like a yes to me.

i wonder what is that amd processor may or may not come out in january... 😗 😀
fingers crossed for a 3M version!!! (yea yea i know... but still!!!)

update: read toms' review on r9 290 and.... amd's reference cooler is crap. amd could, at least, stick a single gpu version of it's [strike]shameless gigabyte windforce 3x ripoff[/strike] reference 7990 cooler on the r9 290x and 290 cards + 2x 8 pin pcie power connectors. a 3rd party cooler shows the potential of the press sample. i don't understand whom were amd trying to fool with 1x 6 pin and 1x 8 pin pcie power connectors and the crap blower. high end enthusiasts rarely bother with gfx card power consumption (or so i hear).
kudos to amd for copying nvidia kepler by crippling dp performance and for copying intel's haswell by supplying reviewers with golden samples. the student has become the master. mwa.ha.ha.ha. (actually both are long running industry practices that fanboys are blind to 😀)

 

[juanrga]

Let us forget that pclab is not neutral regarding AMD vs Intel, even accepting your non-marketing benchmark

trinity != richland

athlon != APU

GTX 780 != R9 280X

...

I would like to see multiplayer benchmarks in the same conditions that the benchmark in the slide #13. That is, I would like to see benchmarks of the new APU with the new APU drivers, the new APU optimized Windows, 2133 MHz ram, and paired with a R9 280X...

I estimate it would obtain ~70% of the performance of 8350.

Actually, Richland and Trinity are the same performance wise (same uArch). They didn't move anything or improved anything in regards to speed ups, just efficiency AFAIK.

And the Athlon II 750K IS a de-iGPU'ed APU. No L3 cache and the same speed with a little than you'd-like TDP because of the iGPU space. So comparing it with APUs is fair IMO. As long as the iGPU is not being used, of course.

Cheers!

Yes, same uArch, but Richland introduces several TDP advantages including new temperature sensors installed on the die and an improved frequency adjustment algorithm. This improved efficiency that you correctly note, allows for an increase in base frequencies and more aggressive Turbo Core technology (*). Richland APU is about 5-10% faster than Trinity Athlon

http://www.cpu-world.com/Compare/176/AMD_A10-Series_A10-6800K_vs_AMD_Athlon_X4_750K.html

The real strength of a APU+dGPU configuration is to use the iGPU in tandem with the dGPU. You cannot do that with an Athlon + Nvidia GTX. Recall than modern Athlons are APUs with a defective iGPU. It is disabled because doesn't work and AMD can still reuse the CPU.

Adds new APU drivers and OS

http://semiaccurate.com/2013/10/22/windows-8-1-makes-amds-apus-faster/

and adds 2133 MHz memory and the new APU will be significantly faster than the old Athlon.

(*) There are other small differences like the top Richland introducing native support for 2133MHz memory.

 

[juanrga]

New Steam Machine details revealed
http://techreport.com/news/25606/new-steam-machine-details-revealed

Valve is apparently "super interested" in Intel's Iris Pro IGP—and when asked if Intel or AMD could make processors with beefed-up integrated graphics specifically for Steam Machines, the firm said, "We don't have an answer that we can give you before January." That sure sounds like a yes to me.

i wonder what is that amd processor may or may not come out in january... 😗 😀
fingers crossed for a 3M version!!! (yea yea i know... but still!!!)

It is not apparent. Valve is really interested in integrated graphics, just as the rest of the game industry. The original is here:

There will be a number of different Steam Machine boxes on sale in 2014, and Valve expects them to arrive mid-year. Some of those boxes will be far smaller and / or cheaper than Valve's own prototype unit. "You can get far smaller, and that's what many OEMs are doing… I think it's safe to say less than a quarter of the size," the team told me.

Of course, when they get to that size, they won’t be using full-size graphics cards any more. Intel's integrated graphics are a possibility there: "We're super interested in Iris Pro." When I ask whether Intel or AMD might make special chips to bring down the price of truly powerful integrated graphics, the room goes quiet for a moment. "We don’t have an answer that we can give you before January," the team answers.

There is no 3M Steamroller or it would be used in Berlin APU/CPU. However, Valve is working with custom Nvidia hardware according to details of the SteamBox prototype. This opens the door for AMD custom solutions dept. to design something exclusive for the Steambox. AMD could, if Valve want to pay for that, design a 8-core Kaveri APU by combining two quad-core dies (similar to Xbox1 and PS4 custom APUs).

 

[juanrga]

In a completely unrelated topic, this thread is a year old today so cheers

Just in time to cite this

http://www.brightsideofnews.com/news/2013/11/4/what-to-expect-from-kaveri-a-detailed-predictive-analysis.aspx

 

[noob2222]

Spoiler

The slide shows the APU giving ~92% of the FX-8350 performance in that gaming test. Say us which will be the performance in "multiplayer"?

athlon x4 750k = trinity. its found at the bottom by the pentium g's, overclocked to 4.5 ghz and still getting beat by 4350 @ 4.2 ghz and completely obliterated by the 8350.

you really need to quit trying to live by marketing slides. best case scenario != all cases.

While (Juanrga = wrong) {printf("JUANRGA IS NEVER WRONG")};

Let us forget that pclab is not neutral regarding AMD vs Intel, even accepting your non-marketing benchmark

trinity != richland

athlon != APU

GTX 780 != R9 280X

...

I would like to see multiplayer benchmarks in the same conditions that the benchmark in the slide #13. That is, I would like to see benchmarks of the new APU with the new APU drivers, the new APU optimized Windows, 2133 MHz ram, and paired with a R9 280X...

I estimate it would obtain ~70% of the performance of 8350.

so basically what your getting at is you only want to see kaveri running highly specialized software to take advantage of HSA and ignore any programs that are in the market today.

This is because in the programs of today, Kaveri will suck hind teat.

Just in time to cite this

http://www.brightsideofnews.com/news/2013/11/4/what-to-...

rofl, juan got his blog site quoted. I wonder how bad BSN will look when it turns out not even remotely accurate.

 

[de5_Roy]

Just in time to cite this
http://www.brightsideofnews.com/news/2013/11/4/what-to-expect-from-kaveri-a-detailed-predictive-analysis.aspx

congratulations, juanrga. however, this event only raises my expectation of you being able to properly interpret marketing, promotions, speculations, rumors, benchmarketing, benchmarks and common industry practices. and to be able to correctly distinguish them. i wish you good luck.

 

[gamerk316]

AMD needs market share. With that the money will start flowing into their R&D (i hope) and the CEO yatches and beach houses, haha. Probably a big chunk will go to the later, lol.

Market share measures past sales. Its nothing more then an indication of how competitive ones products are. Gaining share at the expense of profit is pointless, however. No profit = No money to go into R&D, which leads to no product and less market share in the future.

Get it?

In a completely unrelated topic, this thread is a year old today so cheers

Just in time to cite this

http://www.brightsideofnews.com/news/2013/11/4/what-to-expect-from-kaveri-a-detailed-predictive-analysis.aspx

The following are all CPU performance estimates

 

[tracker45]

Who uses an APU for gaming ?? No serious gamer does.

 

[tracker45]

Steamroller is looking more and more disappointing with each article. It looks like it performs like the current mid range i5 CPUs.

 

[Cazalan]

If you want to get creative with marketing slides you should take those "benchmarks" and add sliders to them, so you can turn the % gain from 5-30% and then adjust the base/turbo frequencies from 3-4Ghz.

 

[Cazalan]

Steamroller is looking more and more disappointing with each article. It looks like it performs like the current mid range i5 CPUs.

If the price is around the same or less than Richland APUs ($139) that would be a solid win for AMD. i5's are in the $179-239 range.

 

[gamerk316]

Steamroller is looking more and more disappointing with each article. It looks like it performs like the current mid range i5 CPUs.

Well, considering everyone has been proclaiming that SR would be around 2500k levels, which is about what a mid-tier i5 does, I fail to see how that's surprising. Its what everyone has been predicting for about a year now.