AMD CPU speculation... and expert conjecture

[jed]

And the 3930x costs 5x what the 8350 does...

Somebody needs to track down BaronMatrix. I don't know if there is enough popcorn in the States to sustain me for the flame war these two could have.

No it's not 5x more then the FX 8350 it's 2 1/2 time the cost of a 8350, and
the FX 9590 is over price at 899.00 for the performance you will receive,
http://www.newegg.com/Product/Product.aspx?Item=N82E16819113347
when a 3930 will perform much better and if overclocked the FX 9590 can't
touch it for the lower cost of 569.00.

http://www.newegg.com/Product/Product.aspx?Item=N82E16819116492

http://www.newegg.com/Product/Product.aspx?Item=N82E16819113284

 

[etayorius]

I took the easy exit. Hopefully action will be taken soon.

I can't believe I had to read almost 2 pages of trolling just to read... 0 news about Steamroller and 2 constructive posts in this sea of floating manure being posted (the bait and the baited, I might add).

STOP. FEEDING. THE. TROLL.

Cheers!

Yes please just ignore this dude, he is trying his best to hijack this Thread, it is getting annoying how YOU PEOPLE keep up with his SHIT, i came here for AMD and Steamroller news and the last "F" thing i want to read is how supreme, innovative or how many inches Intel "D" is, Haffijur seems to know the exact lenght of that thing...

Please just ignore him, DO NOT REPLY TO HIS comments any further.


Kinda reminds me of the JML12 that is all over in Most Tech news sites with his crusade against AMD.

 

[rmpumper]

Comparing the i7 4770K is already a stretch at over $300. If Steamroller has 20-30% increase over Piledriver it will be neck and neck with i7 4770K which is still almost double the cost.

You are delusional if you think that AMD would be selling 4770K equivalent Steamroller for the price of 8350. 8350 is priced as it is because it performs on average as an i5.

 

[sarinaide]

I am on about actual cpu performance. Look here the i7 3960x blows the fx9590 away. Even at stock it would. The i7 4770k 3770k or 2600k is the fx9590 competitor and we all know how if you oc these cpus past 3.5ghz it will destroy the fx9590 which is already at the limit.
http://www.kitguru.net/components/cpu/zardon/amd-fx9590-5ghz-review-w-gigabyte-990fxa-ud5/14/

edit: If you compare 32nm vs 32nm the i7 2700k vs the fx9590 is what you should be looking at as thats its performance category and we all know how well sandy bridge can oc and at 5ghz it will destroy the fx9590 and spit it out for breakfast.

http://www.kitguru.net/components/cpu/zardon/amd-fx9590-5ghz-review-w-gigabyte-990fxa-ud5/25/

Since you so concerned about Power and Overclocks on Intel side there is your 3960X at 4.8ghz using over 100w more to burn, kudos on making yourself look a dunce, maybe you should find better benches.

 

[noob2222]

so what. your oc sb-e laptop is only going to last 30 minutes or less.

 

[mlscrow]

Thats because it was heavily oc, at stock it already beats it by 50% at much lower power, oc it is close to double the performance.

Are you seriously an idiot or just a troll? Please answer, then shut up.

 

[juanrga]

http://www.anandtech.com/bench/Product/362?vs=697
krnt that was me. There are of course different stuff on the piledriver cpu architecturally but if you look at these benchmarks you could say a phenom 2 8 core would beat the fx8350, suprisingly enough phenom 2 cpus at same clock speed as piledriver it seems most likely the older cpu has better ipc and people think intel are not innovating. I think thats why intel must have been happy knowing amd have stood still for last 4 years in desktop market except a die shrink to 32nm while sandy bridge was so good the extreme line cpu was worth the prices as nothing like that is out in the market and it was for sure amd wouldn't bother intel competition wise while they sticked with bulldozer and piledriver. To be fair I don't think amd expected intel to innovate so much since the junk nehalem 45nm cpus came out, sandy bridge shocked me when the benchmarks came out especially the laptop cpus I remember reading the benchmarks in january 2011 and thinking wow. AMD need a wow type amazing cpu to get back in the game.

TROLLING ALGORITHM:

while ( 3 > 1 )
{
  post nonsense;
  ignore corrections;
}

 

[hcl123]

Intel is not using its power efficiency advantage till 8 core haswell. Intel don't need to improve drastically as they are already on a sensational level. Its true that amd need 8 core to compete against intel 4 cores.

Ridiculous... the differences in performance ARE RIDICULOUS.

I've shown you from HPC world how to be possible to have 10x more performance out of the same metal at the same speed, and you talk about 10 to 20% difference in "windows world" as something "sensational" ? ... don't be ridiculous also...

Besides you are all mixed up, in REAL WORLD, multithreading by 8 cores, even if they are somehow restricted, will always win for 4. The problem is that windows world doesn't have those much "threads", Intel doesn't need 8 real cores when software is not even 4 threads generalized.... IPC is plateaued, bulk is not prone to high clocks, though in this clock department, things are also not going much higher under the current FAB processes... only rests to Intel... and AMD for that matter... to rely on "propaganda" to sell OLD deigns as something sensational LOL (are you part of it ?)

If i'm campaigning, i'm campaigning for ppl to see if they want an "evolution" FIRST they must ask for more thread software, more, better, more performant software... the current paradigma is in a dead end... and without wanting to be biased i think we can thank a lot intel with their single-thread mantra for the situation, Windows client/desktop will be fine with 4 cores for a lot of time to come, there is no "incentive" to "evolve" in that world.

Real disgusting is to see the marketing departments "invent" new tricks to push the same old same... soon self appointed morons or payed posters will fill forums with posts evangelizing 2 or 3% differences as something of the other world LOL... no wonder ARM is winning, the Windows Traditional Desktop is in a self inflicted paralysis.

But there is hope... if the "powers" of the traditional PC world don't evolve, perhaps ARM will, perhaps a major mass change can be brewing in the cauldron of events.

 

[Fasting]

ASUS announces two FM2+ motherboards: http://translate.google.se/translate?sl=sv&tl=en&js=n&prev=_t&hl=sv&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fnyhet%2F17328-asus-tillkannager-fm2-moderkort-med-stod-for-amd-a-serien-kaveri

• 
  Asus A88XM-A (Apparently named "Slurry" by Google )
  Asus A55BM-A/USB3

 

[Cazalan]

I thought after the quiet period of the earnings report we'd get some news.

 

[BeastLeeX]

ASUS announces two FM2+ motherboards: http://translate.google.se/translate?sl=sv&tl=en&js=n&prev=_t&hl=sv&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fnyhet%2F17328-asus-tillkannager-fm2-moderkort-med-stod-for-amd-a-serien-kaveri

• 
  Asus A88XM-A (Apparently named "Slurry" by Google )
  Asus A55BM-A/USB3

No DDR4!

 

[blackkstar]

seeing as x264 is open source, does that mean their version is compiled with ICC? such that intel cpus use the avx2 and fma support while AMD is reduced to sse2?
http://forum.doom9.org/showthread.php?t=158934

Id say the chances are high.

I benchmarked between x264 in Windows and Gentoo (with bdver2 CFLAGs) and Gentoo was about 10% faster than in Windows. Just because it is open source, does not mean it is a good benchmark. Open source benchmarks, however, are good because you can recompile it how you like and test how fair the reviewer is being.

 

[juanrga]

ASUS announces two FM2+ motherboards: http://translate.google.se/translate?sl=sv&tl=en&js=n&prev=_t&hl=sv&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fnyhet%2F17328-asus-tillkannager-fm2-moderkort-med-stod-for-amd-a-serien-kaveri

• 
  Asus A88XM-A (Apparently named "Slurry" by Google )
  Asus A55BM-A/USB3

No DDR4!

We already knew kaveri comes with DDR3.

DDR4 is not ready for 2013 and I don't see real advantages, at least initially:

Initial DDR4 will be expensive and performance will be the same than DDR3 (we already have 2600 and 2400 DDR3 memory). Only at later years (2016?) DDR4 will start to show a performance boost over DDR3. Recall kaveri will be replaced in 2015. Therefore DDR4 support does not make many sense for me.

Moreover, to get full performance, you will need a DIMM-DDR4 in every slot on the board due to simplification of topology.

In any case what is the promise of 50GB DDR4 when compared with GDDR5 already surpassing the 200GB?

I still have hopes to see GDDR5 support in some kaveri model, but it is unlikely.

 

[blackkstar]

While I'm here I thought I would share something with Hajifur.

http://www.bestofmedia.com/audience.php?langue=en

You can buy advertising for Intel directly from Tom's Hardware's parent company. You should just finally give up and stop wasting your time and instead just pay to advertise for Intel and walk away.

I will say, though, that people like you are a great way to sell AMD chips. I simply mention how ridiculous you are about mentioning things that are irrelevant (power consumption, comparing more expensive Intel chips to cheaper AMD ones and saying Intel is better, Cinebench, etc), apply a broad brush to the average Intel fanboy and then Intel fanboys look like shilling retards.

 

[cowboy44mag]

While I'm here I thought I would share something with Hajifur.

http://www.bestofmedia.com/audience.php?langue=en

You can buy advertising for Intel directly from Tom's Hardware's parent company. You should just finally give up and stop wasting your time and instead just pay to advertise for Intel and walk away.

I will say, though, that people like you are a great way to sell AMD chips. I simply mention how ridiculous you are about mentioning things that are irrelevant (power consumption, comparing more expensive Intel chips to cheaper AMD ones and saying Intel is better, Cinebench, etc), apply a broad brush to the average Intel fanboy and then Intel fanboys look like shilling retards.

+1 :lol:

Seriously though, the troll has been quiet for awhile (power outage maybe?:lol, lets not poke him with a stick

 

[cowboy44mag]

ASUS announces two FM2+ motherboards: http://translate.google.se/translate?sl=sv&tl=en&js=n&prev=_t&hl=sv&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fnyhet%2F17328-asus-tillkannager-fm2-moderkort-med-stod-for-amd-a-serien-kaveri

• 
  Asus A88XM-A (Apparently named "Slurry" by Google )
  Asus A55BM-A/USB3

No DDR4!

We already knew kaveri comes with DDR3.

DDR4 is not ready for 2013 and I don't see real advantages, at least initially:

Initial DDR4 will be expensive and performance will be the same than DDR3 (we already have 2600 and 2400 DDR3 memory). Only at later years (2016?) DDR4 will start to show a performance boost over DDR3. Recall kaveri will be replaced in 2015. Therefore DDR4 support does not make many sense for me.

Moreover, to get full performance, you will need a DIMM-DDR4 in every slot on the board due to simplification of topology.

In any case what is the promise of 50GB DDR4 when compared with GDDR5 already surpassing the 200GB?

I still have hopes to see GDDR5 support in some kaveri model, but it is unlikely.

While I would love to see Kaveri with DDR5, I don't think its going to happen. AMD is going to want Kaveri to be cost competitive above all, and DDR4 or DDR5 would just make the total system builds too high for not that much gain (yet). Hopefully Excavator will have DDR5

 

[juanrga]

I benchmarked between x264 in Windows and Gentoo (with bdver2 CFLAGs) and Gentoo was about 10% faster than in Windows. Just because it is open source, does not mean it is a good benchmark. Open source benchmarks, however, are good because you can recompile it how you like and test how fair the reviewer is being.

bdver2 on GCC doesn't extract all the performance form piledriver arch. However FX-8350 is already slightly faster than i7-3770k on x264 with bdver2 on GCC 4.7:

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

The score of the 8350 @ 4.6GHz gives an idea of the performance of the FX-9590 (4.7GHz).

Kaveri (assuming 30% over trinity) would score between the i5-2500k and the i5-3470.

I am preparing a series of graphs with estimated performance of Steamroller cores in kaveri.

P.S: I am interested in Gentoo. Can I PM you with a pair of doubts?

 

[juanrga]

ASUS announces two FM2+ motherboards: http://translate.google.se/translate?sl=sv&tl=en&js=n&prev=_t&hl=sv&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fnyhet%2F17328-asus-tillkannager-fm2-moderkort-med-stod-for-amd-a-serien-kaveri

• 
  Asus A88XM-A (Apparently named "Slurry" by Google )
  Asus A55BM-A/USB3

No DDR4!

We already knew kaveri comes with DDR3.

DDR4 is not ready for 2013 and I don't see real advantages, at least initially:

Initial DDR4 will be expensive and performance will be the same than DDR3 (we already have 2600 and 2400 DDR3 memory). Only at later years (2016?) DDR4 will start to show a performance boost over DDR3. Recall kaveri will be replaced in 2015. Therefore DDR4 support does not make many sense for me.

Moreover, to get full performance, you will need a DIMM-DDR4 in every slot on the board due to simplification of topology.

In any case what is the promise of 50GB DDR4 when compared with GDDR5 already surpassing the 200GB?

I still have hopes to see GDDR5 support in some kaveri model, but it is unlikely.

While I would love to see Kaveri with DDR5, I don't think its going to happen. AMD is going to want Kaveri to be cost competitive above all, and DDR4 or DDR5 would just make the total system builds too high for not that much gain (yet). Hopefully Excavator will have DDR5

I shared my hopes about GDDR5.

DDR5 doesn't exist.

 

[noob2222]

ASUS announces two FM2+ motherboards: http://translate.google.se/translate?sl=sv&tl=en&js=n&prev=_t&hl=sv&ie=UTF-8&u=http%3A%2F%2Fwww.sweclockers.com%2Fnyhet%2F17328-asus-tillkannager-fm2-moderkort-med-stod-for-amd-a-serien-kaveri

• 
  Asus A88XM-A (Apparently named "Slurry" by Google )
  Asus A55BM-A/USB3

No DDR4!

We already knew kaveri comes with DDR3.

DDR4 is not ready for 2013 and I don't see real advantages, at least initially:

Initial DDR4 will be expensive and performance will be the same than DDR3 (we already have 2600 and 2400 DDR3 memory). Only at later years (2016?) DDR4 will start to show a performance boost over DDR3. Recall kaveri will be replaced in 2015. Therefore DDR4 support does not make many sense for me.

Moreover, to get full performance, you will need a DIMM-DDR4 in every slot on the board due to simplification of topology.

In any case what is the promise of 50GB DDR4 when compared with GDDR5 already surpassing the 200GB?

I still have hopes to see GDDR5 support in some kaveri model, but it is unlikely.

DDR4 will not be a rush-to get transition. It will be similar to ddr2 to ddr3. Latency will be horrid and speeds will start where ddr3 is leaving off.

http://www.micron.com/products/dram/ddr4-sdram#fullPart&236=0&233=1

cl-13 for ddr4 1866 and cl-14 for 2133. ddr4 will initially benefit servers more than end-users.

I benchmarked between x264 in Windows and Gentoo (with bdver2 CFLAGs) and Gentoo was about 10% faster than in Windows. Just because it is open source, does not mean it is a good benchmark. Open source benchmarks, however, are good because you can recompile it how you like and test how fair the reviewer is being.

I always question a review's intention when they start to include "custom designed" benchmarks. Custom designed for who?

 

[szatkus]

I benchmarked between x264 in Windows and Gentoo (with bdver2 CFLAGs) and Gentoo was about 10% faster than in Windows. Just because it is open source, does not mean it is a good benchmark. Open source benchmarks, however, are good because you can recompile it how you like and test how fair the reviewer is being.

bdver2 on GCC doesn't extract all the performance form piledriver arch. However FX-8350 is already slightly faster than i7-3770k on x264 with bdver2 on GCC 4.7:

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

The score of the 8350 @ 4.6GHz gives an idea of the performance of the FX-9590 (4.7GHz).

Kaveri (assuming 30% over trinity) would score between the i5-2500k and the i5-3470.

I am preparing a series of graphs with estimated performance of Steamroller cores in kaveri.

P.S: I am interested in Gentoo. Can I PM you with a pair of doubts?

GCC is the best compiler for Piledriver and Steamroller. BdverX was prepared by AMD (open source ).

 

[Ranth]

http://www.asus.com/Motherboards/A88XMA/#specifications

Seems like AMD got around to get PCIe3.0

"1 x PCIe 3.0/2.0 x16 *2
*2: Only FM2+ Processors can support PCIe3.0."

 

[jdwii]

Juangra I am a troll when you spout false information. Kaveri if 30% better would be in the middle of an i3 and an i5.

If performance per watt is the most important thing then why are we even using a x86 design?

 

[szatkus]

http://www.asus.com/Motherboards/A88XMA/#specifications

Seems like AMD got around to get PCIe3.0

"1 x PCIe 3.0/2.0 x16 *2
*2: Only FM2+ Processors can support PCIe3.0."

There was news about it. This mobo is on the market?

Juangra I am a troll when you spout false information. Kaveri if 30% better would be in the middle of an i3 and an i5.

You both trolling

 

[8350rocks]

Juangra I am a troll when you spout false information. Kaveri if 30% better would be in the middle of an i3 and an i5.

 

[szatkus]

Juangra I am a troll when you spout false information. Kaveri if 30% better would be in the middle of an i3 and an i5.

(some benches...)

We're not talking about GPU. Kaveri's GPU will beat even HD 5200. It's about CPU.