AMD CPU speculation... and expert conjecture

[etayorius]

AMD will disclose Carrizo, an integrated processor with its latest x86 core. The 28 nm chip measures 244.62 mm2 and packs more than 3.1 billion transistors. Its new Excavator core is 23% smaller and uses 40% less power than AMD’s previous x86 core.

Can we start FD-SOI rumors again?

Can`t seem to find who did you quoted that from.

 

[szatkus]

AMD will disclose Carrizo, an integrated processor with its latest x86 core. The 28 nm chip measures 244.62 mm2 and packs more than 3.1 billion transistors. Its new Excavator core is 23% smaller and uses 40% less power than AMD’s previous x86 core.

Can we start FD-SOI rumors again?

Can`t seem to find who did you quoted that from.

Really? Google showed the source on first page, when I pasted it.

Anyway, source: http://www.eetimes.com/document.asp?doc_id=1324643&page_number=2

 

[juanrga]

AMD will disclose Carrizo, an integrated processor with its latest x86 core. The 28 nm chip measures 244.62 mm2 and packs more than 3.1 billion transistors. Its new Excavator core is 23% smaller and uses 40% less power than AMD’s previous x86 core.

Can`t seem to find who did you quoted that from.

The same quote was given yesterday by SeronX here. SeronX gives the link to the source for the quote and gave the thanks by the info to someone whom I know very well. The same quote has been now reproduced in many forums and news sites.

 

[logainofhades]

Real or fake, with regards to Ubisoft, they screwed up. I work in quality control, so the get it right the first time is pretty much ingrained into me.

 

[gamerk316]

Real or fake, with regards to Ubisoft, they screwed up. I work in quality control, so the get it right the first time is pretty much ingrained into me.

Trust me, when deadlines are involved, it's all about "risk management". And that isn't limited to gaming either.

 

[logainofhades]

Yea, I have seen engineers cut corners, just for it to blow up in their face later. Hence why quality and product engineers seem to butt heads at times, because it is usually the quality engineer that has to deal with the customer, when things go wrong.

 

[8350rocks]

@logain: there is a great deal of truth there...though, I would say software engineers tend to do that more often. In terms of hardware, especially microprocessor design, cutting corners is a bit harder. Especially on more advanced processes. Though some recent examples of that would include the shoddy thermal solution used on intel desktop cpus. I am sure there are more examples...that one just comes to mind.

 

[colinp]

Phenom I?

 

[8350rocks]

Phenom I?

I think that was an oversight, and not an intentional corner being cut. Meanwhile, Intel did not "accidentally" put cheap thermal solution in chips aimed at overclocking...

 

[-Fran-]

Phenom I?

I think that was an oversight, and not an intentional corner being cut. Meanwhile, Intel did not "accidentally" put cheap thermal solution in chips aimed at overclocking...

8150sucks?

Sorry, had to do it.

Cheers! XD

 

[szatkus]

Real or fake, with regards to Ubisoft, they screwed up. I work in quality control, so the get it right the first time is pretty much ingrained into me.

Trust me, when deadlines are involved, it's all about "risk management". And that isn't limited to gaming either.

Phenom I?

I think that was an oversight, and not an intentional corner being cut. Meanwhile, Intel did not "accidentally" put cheap thermal solution in chips aimed at overclocking...

Evil Intel. Oh, wait...

http://www.hardcoreware.net/amd-kaveri-delidded/

 

[colinp]

It was a failure in quality control, so the testing phase may not have been as complete as it should have been. Certainly, Phenom did not perform as expected. Arguably, the Intel chips did perform as expected out of the box, just not quite to the elevated levels of expectation from previous generations.

I always felt that Bulldozer was forced out of the door in an incomplete state too. It was so overdue that to have delayed any further would have spooked investors more than the underwhelming reception it received. Or so the reasoning would have gone internally.

 

[8350rocks]

@szatkus:

Were that an FX chip...ok. Though I think no one here has any illusions about kaveri and the market segment it is targeted toward. Meanwhile a 4670k is not in the same bracket.

@colinp:

Now, I think bulldozer was not rushed per se, I think the architecture has a higher theoretical ceiling than AMD realized and the single thread bottlenecks were greater than anticipated. I also think that they failed to realize how far short of theoretical maximums it would fall in single thread. Multithread, it is still aging well.

 

[jdwii]

http://international.download.nvidia.com/geforce-com/international/comparisons/far-cry-4/far-cry-4-texture-quality-comparison-3-ultra-vs-low.html

Am i the only one who thinks Low textures looks extremely like ultra, and in some cases better on low like the one below.
http://international.download.nvidia.com/geforce-com/international/comparisons/far-cry-4/far-cry-4-texture-quality-comparison-2-ultra-vs-low.html

Been running far cry 4 at ultra except i have textures on medium, this game is a lot of fun i hope you Amd guys can have fun with it soon i will say it works great on the 8350.

 

[Reepca]

Being a rather-newbie Java programmer, I was curious how I could apply HSA to any programs I may want to write in the future. I checked out aparapi, and tried to build the HSA emulator, then to build okra to interface with it, and 3 days later... still no sign of getting anywhere. For an up-to-date ubuntu installation, the build process has found a remarkable number of ways to screw up. Unfortunately either there aren't any people out there getting the same errors as me, they all are naturally knowledgeable enough to come up with their own solutions, or they simply don't ask questions online.

Seems like everyone interested in HSA has a solid understanding of how every piece of it works... except me. Conceptual understanding is easy, but it won't make code from another language you don't understand compile. Perhaps I should just wait for the software to become more polished.

(End emotional release).

 

[colinp]

@szatkus:

Were that an FX chip...ok. Though I think no one here has any illusions about kaveri and the market segment it is targeted toward. Meanwhile a 4670k is not in the same bracket.

@colinp:

Now, I think bulldozer was not rushed per se, I think the architecture has a higherntheoretical ceiling than AMD realized and the single thread bottlenecks were greater than anticipated. I also think that they failed to realize how far short of theoretical maximums it would fall in single thread. Multithread, it is still aging well.

Wow double standards, much. Two chips, both with unlocked multipliers, both with TIM instead of solder. But one is Intel (evil) one is AMD (can do no wrong, ever).

 

[juanrga]

http://ir.amd.com/phoenix.zhtml?c=74093&p=RssLanding&cat=news&id=1989919

AMD will conduct research for an integrated exascale node architecture based on its HSA-enabled APUs. Particular areas of research include power efficiency, reliability, programmability, component and network interface integration, APU microarchitecture, advanced memory architectures and efficient data movement.

AMD will collaborate with the DOE and others to help define a new standard for memory interfaces that meets the needs of future-generation memory devices, including non-volatile memory and processing-in-memory (PIM) architectures.

According to Alan Lee, AMD's corporate vice president for Research and Advanced Development:

This research is focused on energy-efficient node architectures and memory systems to improve the capability of the world's fastest supercomputers.

Those future designs are aimed to:

power a generation of exascale supercomputers capable of delivering 30--60 times more performance than today's fastest supercomputers.

 

[8350rocks]

@juanrga:

Not APUs alone in those rigs. Most certainly not...power consumption would be the first issue, cpu interconnects the second. Sure they can design a new interface, but why would you run a ton of 1 gflop pieces at 95w each over a 5+ tflop part for only 3 times the power draw?

Answer: you would not. No matter how far APUs come, they will not be the sole component. I am not going to go into detail, but let us say my thoughts are confirmed.

 

[logainofhades]

Phenom I?

I think that was an oversight, and not an intentional corner being cut. Meanwhile, Intel did not "accidentally" put cheap thermal solution in chips aimed at overclocking...

You mean like the Athlon 750k/760k? Or the crappy cooler bundled with the FX 6300? AMD bundles all their less than 100w chips with a crappy all aluminum heatsink. Both companies are guilty of this one.

 

[8350rocks]

Phenom I?

I think that was an oversight, and not an intentional corner being cut. Meanwhile, Intel did not "accidentally" put cheap thermal solution in chips aimed at overclocking...

You mean like the Athlon 750k/760k? Or the crappy cooler bundled with the FX 6300? AMD bundles all their less than 100w chips with a crappy all aluminum heatsink. Both companies are guilty of this one.

Yes, the aluminum cooler is an example of cutting corners...though, in all fairness...that chip does only cost $100 and is not a flagship part either. However, yes, the under 100W parts got a crappy cooler. (No one uses the stock cooler anyway do they?)

 

[8350rocks]

@szatkus:

Were that an FX chip...ok. Though I think no one here has any illusions about kaveri and the market segment it is targeted toward. Meanwhile a 4670k is not in the same bracket.

@colinp:

Now, I think bulldozer was not rushed per se, I think the architecture has a higherntheoretical ceiling than AMD realized and the single thread bottlenecks were greater than anticipated. I also think that they failed to realize how far short of theoretical maximums it would fall in single thread. Multithread, it is still aging well.

Wow double standards, much. Two chips, both with unlocked multipliers, both with TIM instead of solder. But one is Intel (evil) one is AMD (can do no wrong, ever).

4670k != 7850k

If we were discussing this in terms of space, the 7850k would be earth in the milky way. The 4670k would be a light source trillions of light years off in the distance.

This is not a double standard. The FX8350 is a more apt comparison to that chip, and it does not use TIM. The 7850k is not a HEDT chip. To be brutally honest, it is an OEM chip for DT PCs in $500-800 prebuilt rigs. It is also a decent HTPC processor to use without a dGPU. Though if you look at benchmarks, the 7850k is not in the same ballpark as the 8350. Nor was it ever intended to be. APUs are not HEDT chips at this point.

 

[szatkus]

(No one uses the stock cooler anyway do they?)

Yup, only 99% people I know

 

[szatkus]

4670k != 7850k

AMD marketing wouldn't agree with you.

http://cdn.wccftech.com/wp-content/uploads/2014/01/A10-7850K-vs-Core-i5-4670K1.jpg

Hell, it was even just slightly cheaper than i5 at the begining.

 

[gamerk316]

http://www.extremetech.com/computing/194495-intels-third-generation-xeon-phi-to-use-10nm-technology-deploy-second-generation-omni-path-fabric

 

[gamerk316]

4670k != 7850k

AMD marketing wouldn't agree with you.

http://cdn.wccftech.com/wp-content/uploads/2014/01/A10-7850K-vs-Core-i5-4670K1.jpg

Hell, it was even just slightly cheaper than i5 at the begining.

So wait a sec, when did Intel catch up in IGP performance? That graph makes me LESS INCLINED to buy AMD, since we know the CPU portion of the chip is much stronger on the Intel side.