First Steamroller processor core exposure

[Abwx]

Please provide link to your source(other than yourself), because in reality FX-8350 is nowhere near 3960X, 4670K or 3570K, and we yet skipped 3770 and 4770K.
And also explain how cinebench is "Intel-only opimized" and explain which benchmarks are "neutral".

This test measures CPU performance using the SHA1 hashing algorithm defined in the Federal Information Processing Standards Publication 180-3. The code behind this benchmark method is written in Assembly, and it is optimized for every popular AMD, Intel and VIA processor core variant by utilizing the appropriate MMX, MMX+/SSE, SSE2, SSSE3, or AVX instruction set extension.

http://www.guru3d.com/articles_pages/amd_fx_8350_processor_review,16.html

 

[NostaSeronx]

And also explain how cinebench is "Intel-only opimized"

Cinebench R11.5 utilizes only SSE2 for AMD and SSE4.1/AVX for Intel.

and explain which benchmarks are "neutral".

Any benchmark that allows all competitors to utilize the highest instruction set. Another rule set for benchmarks is that it actually applies to the real world.

http://www.guru3d.com/articles_pages/amd_fx_8350_processor_review,16.html

Also, with the advent of C++ AMP for Windows. The floating point unit's ability to use the Mov instructions, effectively is also a good benchmark for future workloads. Just to note AMD wins with mov engines.

 

[galego]

Please provide link to your source(other than yourself), because in reality FX-8350 is nowhere near 3960X, 4670K or 3570K, and we yet skipped 3770 and 4770K.
And also explain how cinebench is "Intel-only opimized" and explain which benchmarks are "neutral".

Cinebench is one of those biased benchmarks that use the Cripple_AMD function to cheat performance: A CPUID test detects the manufacturer of the CPU, if it is "GenuineIntel" then an optimized code is run, otherwise the chip is forced to run the slowest possible code (even if the AMD chip is compatible with the fastest code).

As stated in my previous post a neutral benchmark is one that optimizes the code for both AMD and Intel and, thus, shows the real performance of each chip. Next, I am giving you some openbenchmarks. "Open" means that the code is open to investigation by anyone.

Here some old examples where the piledriver FX-8350 outperforms the ivy Bridge i7-3770k

http://openbenchmarking.org/embed.php?i=1210227-RA-AMDFX835085&sha=6dc05fb&p=2

http://openbenchmarking.org/embed.php?i=1210227-RA-AMDFX835085&sha=293f200&p=2

http://openbenchmarking.org/embed.php?i=1210227-RA-AMDFX835085&sha=f236ffc&p=2

http://openbenchmarking.org/embed.php?i=1210227-RA-AMDFX835085&sha=5a87e9e&p=2

Here some more recent examples where the FX-8350 outperforms the i7-3960X

http://openbenchmarking.org/embed.php?i=1305170-UT-LLVMCLANG75&sha=78137b2&p=2

http://openbenchmarking.org/embed.php?i=1305170-UT-LLVMCLANG75&sha=82ca41f&p=2

http://openbenchmarking.org/embed.php?i=1305170-UT-LLVMCLANG75&sha=b3c948c&p=2

Of course, there other benchmarks where the Intel chips outperform the FX-8350, but in average the FX-8350 is competitive when the software uses the chip.

Steamroller will be a quantum step over Piledriver. I would consider a 30% performance gain to the above scores.

Something similar for games. Most games are dual/quad threaded and then Intel chips outperform the FX-8350. But when games are more threaded the FX-8350 performs better

Future games will be more threaded because MS and Sony new consoles are 8-core. The FX-8350 is preferred:

We approached a number of developers on and off the record - each of whom has helped to ship multi-million-selling, triple-A titles - asking them whether an Intel or AMD processor offers the best way to future-proof a games PC built in the here and now. Bearing in mind the historical dominance Intel has enjoyed, the results are intriguing - all of them opted for the FX-8350 over the current default enthusiast's choice, the Core i5 3570K.

http://www.eurogamer.net/articles/digitalfoundry-future-proofing-your-pc-for-next-gen

Steamroller will be a beast for gaming. Even considering only a 15% increase in the above score, it will outperform the fastest (and expensive) i7-3970x.

 

[toyota]

lol there you go posting that screenshot again that love when the others show the 8350 getting beat.

"Future games will be more threaded because MS and Sony new consoles are 8-core. The FX-8350 is preferred"

so buy the overall slower processor now in hopes that it might be faster down the road? that was the same ignorant argument when AMD came out with the Phenom 2 X6 and how has that turned out after all those years? we already have games that use more than 2 and even more than 4 threads and in most cases even a freaking i3 beats 8 core Bulldozer. Piledriver and especially Bulldozer will be obsolete quicker than any Ivy i5 or i7 cpu.

so when are you going to buy a pc? I guess you are waiting on the consoles with their magical power that the pc gaming community has yet to ever experience.

And personal attacks are STILL not allowed
-ViRGE

 

[Sheep221]

http://www.guru3d.com/articles_pages/amd_fx_8350_processor_review,16.html

I read the whole review and the FX was behind 5 Intels most of the time.
The single threaded performance is still way behind and gaming performance was limited too.
And then again for much higher power use.
In the case of benchmarking I'm not sure they are intentionally coded to work against AMD. If yes, than why everyone uses them then?

 

[chernobog]

http://www.anandtech.com/show/6201/amd-details-its-3rd-gen-steamroller-architecture

Anandtech has done an overview of Steamroller, of what it should bring.

 

[toyota]

http://www.anandtech.com/show/6201/amd-details-its-3rd-gen-steamroller-architecture

Anandtech has done an overview of Steamroller, of what it should bring.

that info has been out there for nearly a year now...

 

[chernobog]

I know, I made an thread and everyone bashed.

 

[NTMBK]

I know, I made an thread and everyone bashed.

Pointing out gaping holes in your logic != bashing.

 

[NostaSeronx]

Pointing out gaping holes in your logic != bashing.

Actually, he was pretty close to expected Steamroller performance. He just got the clock speeds wrong.

 

[chernobog]

Actually, he was pretty close to expected Steamroller performance. He just got the clock speeds wrong.

Thank you?

So I was correct and only the clock speeds were wrong, how?

Steamroller should improve multi-threaded performance by 35-40% since it has 15-20% IPC/OPS improvements in single/multi-threaded performance and also another 20% in multi-threaded performance since Steamroller does not have multi-threading penalty as Bulldozer/Piledriver.

 

[NTMBK]

So I was close? Ok, then at what clock speeds it should be?

Don't listen to him, he's not an oracle- he's just full of [edit]baloney.

http://forums.anandtech.com/showpost.php?p=34050572&postcount=5

http://forums.anandtech.com/showpost.php?p=34050816&postcount=9

http://forums.anandtech.com/showpost.php?p=34051032&postcount=11

http://forums.anandtech.com/showpost.php?p=34051100&postcount=13

That's all just from one thread...

EDIT: Another gem!

http://forums.anandtech.com/showpost.php?p=34275041&postcount=10

 

[Vesku]

Well equal to or even lower clock than first round Piledriver might explain why AMD is being low key about Kaveri performance... GlobalFoundries hasn't had a great track record with node switch overs especially if you include when it was still part of AMD.

 

[Durp]

What is the current estimated release date for steamroller?

I don't want to support Intel after what they have done with Haswell. (missing instructions on K chips. Purposely limiting overclocks with their gap filled with TIM. Locking down the non-k chips even more) but it's still a MUCH better choice over AMD's current offerings for my usage. (Vishera's horrible IPC and power consumption)

Steamroller needs to be released ASAP to take advantage of this because I doubt I'm the only person who feels this way.

 

[Vesku]

Kaveri is scheduled for late this year. A decent portion of Computex 2013 for AMD was showing that Kaveri was ready. Plans can of course change but I think AMD will try hard to have at least some out in retail by the end of the year.

 

[NostaSeronx]

What is the current estimated release date for steamroller?

November 11-14, 2013.

Well equal to or even lower clock than first round Piledriver might explain why AMD is being low key about Kaveri performance... GlobalFoundries hasn't had a great track record with node switch overs especially if you include when it was still part of AMD.

Kaveri is at TSMC, because Berlin and Kyoto share the same Foundries.

http://forums.anandtech.com/showpost.php?p=34050572&postcount=5

Only the GPU count was off. The fact is Sea Islands which is Enhanced GCN/GCN1.1 is used in both Xbox One and Playstation 4. Steamroller, was based on the slides from AMD, at the time I didn't know the performance of Jaguar. The CPU core count was correct, I don't need to explain GFlops right?

http://forums.anandtech.com/showpost.php?p=34050816&postcount=9

For Excavator, the design was finished but not taped-out. Everything else falls in line with x86-64 emulation of games using PowerPC.

http://forums.anandtech.com/showpost.php?p=34051032&postcount=11

I don't need to explain the first sentence. The second comment is because Jaguar and Steamroller use a type of Ring Bus, which is connected through the L2 SRAM.

http://forums.anandtech.com/showpost.php?p=34051100&postcount=13

IBM was the middle man for all consoles this era: PS4, XB1, WiiU. You will see PS3 ports to PS4 using the same code for the PS3.

http://forums.anandtech.com/showpost.php?p=34275041&postcount=10

Kaveri was initially cancelled but revived later with the new Steamroller revision.

 

[chernobog]

Don't listen to him, he's not an oracle- he's just full of crap.

http://forums.anandtech.com/showpost.php?p=34050572&postcount=5
(old , he was close in the 4 months of 2012 we din't know much about Steamroller, Kaveri nor Jaguar at all. Speculation =/= Full Of Crap)

http://forums.anandtech.com/showpost.php?p=34050816&postcount=9
(AMD could already in 2012 produce first prototype of Excavator for testing, PlayStation 4 is essentially a PC so it can emulate PS1/2 games while emulating PS3 would be a difficult task but possible, the last thing I am not sure)

http://forums.anandtech.com/showpost.php?p=34051032&postcount=11
(They already finished the foundation of Steamroller, the final move is optimization and tweaks thus iterations of the final design while for SPE's being emulated... Only if it was an unified memory)

http://forums.anandtech.com/showpost.php?p=34051100&postcount=13
(ok, this is full of crap)


That's all just from one thread...

Suggestion: Back off fanboy.D:

 

[Durp]

November 11-14, 2013.

Thanks. That's too far off unfortunately. Guess Intel gets at least one build from me this year.

AMD please save us.

 

[Vesku]

November 11-14, 2013.Kaveri is at TSMC, because Berlin and Kyoto share the same Foundries.

That will be embarrassing for GF then, if true, as well as add even more value to the console wins for AMD if they are making both chips at GF to fulfill their WSA.

 

[NTMBK]

Suggestion: Back off fanboy.D:

First one- if it's speculation, say that it is speculation. He stated it as if it were fact.

Second one- yeah, I was just focusing on the middle quote. PS4 has no native backwards compatibility abilities.

Third one- again, I was just looking at the second section. The SPEs had a ridiculous memory model, where you have to manually manage the local memory pools of each of them- Kaveri is basically the exact opposite of that.

How on earth does pointing out that he has a consistent track record of talking nonsense make me a fanboy?

 

[Sheep221]

What is the current estimated release date for steamroller?

I don't want to support Intel after what they have done with Haswell. (missing instructions on K chips. Purposely limiting overclocks with their gap filled with TIM. Locking down the non-k chips even more) but it's still a MUCH better choice over AMD's current offerings for my usage. (Vishera's horrible IPC and power consumption)

Steamroller needs to be released ASAP to take advantage of this because I doubt I'm the only person who feels this way.

I assume they would have to be atleast 20% faster than 2500K and 3770K respectively in order to get existing Intel crowd to upgrade. Just like the 2500K owners didn't upgrade to IVB or Haswell, they won't upgrade to any AMD despite it will be few % faster than fastest Intel. Unless they offer around 20% increase(for them technically at least 40% because they are way behind Intel) they won't be successful with it.
I really wish them luck but it looks unreal to me, as they would really need to redesign this architecture from bottom to achieve that.

How on earth does pointing out that he has a consistent track record of talking nonsense make me a fanboy?

Don't mind that dude, you won't argue with these types because they will bring you down to their level and beat with the experience. His responses are primitive since he joined this forum.

 

[NTMBK]

Kaveri is at TSMC, because Berlin and Kyoto share the same Foundries.

Kaveri is at Global Foundries. http://www.brightsideofnews.com/new...lk-at-gf2c-more-details-from-the-new-wsa.aspx

"If you look at the roadmaps that we have presented at Financial Analyst Day, there is a 28-nanometer successor product to Trinity on the roadmap that we will ramp next year, and that is also manufactured at GLOBALFOUNDRIES."

 

[Vesku]

Kaveri is at Global Foundries. http://www.brightsideofnews.com/new...lk-at-gf2c-more-details-from-the-new-wsa.aspx

Seems to be up in the air:

http://www.brightsideofnews.com/new...aps2c-shows-28nm-kaveri-for-socket-fm22b.aspx

"This allows AMD to potentially make the product at different semiconductor manufacturers. At this point it is unknown whether Kaveri will be made at TSMC or GLOBALFOUNDRIES, but by now GLOBALFOUNDRIES should have their 28nm lines ready for prime time. Previous 28nm products from AMD were exclusively made at TSMC, even though earlier it was announced that in 2013 all processor products would be made at 28nm at GLOBALFOUNDRIES."

We know with Jaguar AMD paid GF so they could have it made at TSMC. I recall in 2012 talk of AMD producing some GPUs on 28nm GF. That seems to have been thrown out as the delays piled on. As we can see GF 28nm has been quite a long time in arriving.

 

[chernobog]

I assume they would have to be atleast 20% faster than 2500K and 3770K respectively in order to get existing Intel crowd to upgrade. Just like the 2500K owners didn't upgrade to IVB or Haswell, they won't upgrade to any AMD despite it will be few % faster than fastest Intel. Unless they offer around 20% increase(for them technically at least 40% because they are way behind Intel) they won't be successful with it.
I really wish them luck but it looks unreal to me, as they would really need to redesign this architecture from bottom to achieve that.
(actually they don't need to redesign from the bottom at all, did Core 2 Duo that was derived from Pentium 3 also a redesign from the bottom? no.)

Don't mind that dude, you won't argue with these types because they will bring you down to their level and beat with the experience. His responses are primitive since he joined this forum.(Keep talking about yourself, it helps )

(We just need to wait and see, either way an FX 45xx will destroy i3, FX 65xx will on the level or slightly beat i5 and FX 85xx will be close or slightly beat an i7, just safe estimation.)

 

[Vesku]

I wouldn't call that kind of prediction "safe estimation". Steamroller SKUs would need to have close to a 40% combined improvement (IPC + Clockspeed + Reduction in CMT penalty) over current Piledriver SKUs to meet those expectations. Perhaps 25-30% if you mean solely in multi-threaded scenarios.