AMD CPU speculation... and expert conjecture

[cowboy44mag]

4 x Jaguar cores = between half and 80% of a Pentium mobile dual core see http://www.anandtech.com/show/6981/the-kabini-deal-can-amd-improve-the-quality-of-mainstream-pcs-with-its-latest-apu/2 I expect SR to be in another league. Jaguar is a competitor to Atom a good one but not good compared to high end desktop CPUs.

Anandtech uses biased benchmarks (Cripple_AMD function) that favour intel chips or uses the ones where AMD chips don't perform so well and ignores the benchmarks where AMD excels. In that review they use pcmark, cinebench, 7-zip...

If you add 33% to the Kabini numbers to simulate the 2Ghz model the results are pretty good. It passes the Pentium in multithreaded benchmarks and still uses less power.

The pentium runs at 2.4GHz and the i5 does up to 2.6GHz. For a clock-clock comparison of raw performance, I would add more than a 33%. Moreover, the effect of the biased benchmarks would be considered as well. It is better to start from fair jaguar benchmarks. Those show that:

4 JG ~ 2 SB + HT
8 JG ~ 4 SB + HT

According to AMD, SR has twice the performance of jaguar; therefore, I think that it is admissible that a 4 SR CPU will be close to an i7-2600k, under fair benchmarks and if everything scales fine.

http://www.techradar.com/reviews/pc-mac/pc-components/processors/amd-a4-5000-apu-1153738/review/3#articleContent

lol, never seen so much BS to make themselves believe amd jaguar cpus can compete with even a sandy bridge cpu with ht. A 2 core high clocked HT intel sandy bridge cpu will beat the 8 core jaguar.

Eh, thats a bit much. 4 core, possibly. 8 core, assuming a benchmark with multiple parallel threads? Not as likely.

+1 2 core Sandy with HT, assuming a benchmark with multiple parallel threads would compete very closely with 4 core Jaguar. To compete against 8 core Jaguar you would need 4 core Sandy with HT. Exactly what juanrga said. Assuming of course a benchmark with multiple parallel threads, now if its single core execution.... Well that's where Steamroller and Kaveri comes in isn't it

 

[szatkus]

Well that's where Steamroller and Kaveri comes in isn't it

No, it's just new generation, not miracle.

 

[juanrga]

4 x Jaguar cores = between half and 80% of a Pentium mobile dual core see http://www.anandtech.com/show/6981/the-kabini-deal-can-amd-improve-the-quality-of-mainstream-pcs-with-its-latest-apu/2 I expect SR to be in another league. Jaguar is a competitor to Atom a good one but not good compared to high end desktop CPUs.

Anandtech uses biased benchmarks (Cripple_AMD function) that favour intel chips or uses the ones where AMD chips don't perform so well and ignores the benchmarks where AMD excels. In that review they use pcmark, cinebench, 7-zip...

If you add 33% to the Kabini numbers to simulate the 2Ghz model the results are pretty good. It passes the Pentium in multithreaded benchmarks and still uses less power.

The pentium runs at 2.4GHz and the i5 does up to 2.6GHz. For a clock-clock comparison of raw performance, I would add more than a 33%. Moreover, the effect of the biased benchmarks would be considered as well. It is better to start from fair jaguar benchmarks. Those show that:

4 JG ~ 2 SB + HT
8 JG ~ 4 SB + HT

According to AMD, SR has twice the performance of jaguar; therefore, I think that it is admissible that a 4 SR CPU will be close to an i7-2600k, under fair benchmarks and if everything scales fine.

http://www.techradar.com/reviews/pc-mac/pc-components/processors/amd-a4-5000-apu-1153738/review/3#articleContent

lol, never seen so much BS to make themselves believe amd jaguar cpus can compete with even a sandy bridge cpu with ht. A 2 core high clocked HT intel sandy bridge cpu will beat the 8 core jaguar.

No beliefs, because the benchmarks showing how a 4 threads jaguar outperform a higher clocked 4 threads SB were given. About your link,

He forgets to mention that the benchmarks that he uses contains the Cripple-AMD function to fake scores. He forgets to mention that the Intel chip has HT and a much higher clock: 2.13 GHz.

I found funny the part where he tries to compare the kabini to FX 4 cores:

A score of 1.45pts is not hugely impressive. That makes the A4-5000 less than half as grunt as a quad-core AMD FX CPU.

Is he aware that the A4 runs at 1.5GHz and lacks turbo? FX-4xxx piledriver have based clocks between 3.8GHz (Fx-4300) and 4.2 GHz (FX-4350), and higher turbo speeds.

In one sense, this confirms my point. Consider a hypothetical FX-4300 downclocked to 1.5 GHz, this would have about the same score than the jaguar. And AMD claims that SR is 2x jaguar. Therefore, this explain why 4C piledrivers are being replaced by 2C SR.

 

[griptwister]

4 x Jaguar cores = between half and 80% of a Pentium mobile dual core see http://www.anandtech.com/show/6981/the-kabini-deal-can-amd-improve-the-quality-of-mainstream-pcs-with-its-latest-apu/2 I expect SR to be in another league. Jaguar is a competitor to Atom a good one but not good compared to high end desktop CPUs.

Anandtech uses biased benchmarks (Cripple_AMD function) that favour intel chips or uses the ones where AMD chips don't perform so well and ignores the benchmarks where AMD excels. In that review they use pcmark, cinebench, 7-zip...

If you add 33% to the Kabini numbers to simulate the 2Ghz model the results are pretty good. It passes the Pentium in multithreaded benchmarks and still uses less power.

The pentium runs at 2.4GHz and the i5 does up to 2.6GHz. For a clock-clock comparison of raw performance, I would add more than a 33%. Moreover, the effect of the biased benchmarks would be considered as well. It is better to start from fair jaguar benchmarks. Those show that:

4 JG ~ 2 SB + HT
8 JG ~ 4 SB + HT

According to AMD, SR has twice the performance of jaguar; therefore, I think that it is admissible that a 4 SR CPU will be close to an i7-2600k, under fair benchmarks and if everything scales fine.

http://www.techradar.com/reviews/pc-mac/pc-components/processors/amd-a4-5000-apu-1153738/review/3#articleContent

lol, never seen so much BS to make themselves believe amd jaguar cpus can compete with even a sandy bridge cpu with ht. A 2 core high clocked HT intel sandy bridge cpu will beat the 8 core jaguar.

Correct me if I am wrong, but doesn't a A4 beat an i3 in performance per watt since its core performance is similar to an i3 and the GPU (or iGPU, what ever you consider it) is better? Trolololo!

*Edit* And let's not forget it's an "Underclocked" CPU. I'd imagine a 8 core Kabini would most definitely crush an i3 in benchmarks with out this anti-AMD code function.

*Edit Again* Since there will be 8 core optimization for the next gen games an FX 8350/8320 should fair just as good or even better than an i7. So does this mean that mean we paid less for the same performance? And more, physical cores?

 

[griptwister]

a 2core ht SB cpu is compareable to a 4 core jaguar.

Lol, The power consumption man! That's what it's all about!

 

[szatkus]

In one sense, this confirms my point. Consider a hypothetical FX-4300 downclocked to 1.5 GHz, this would have about the same score than the jaguar. And AMD claims that SR is 2x jaguar. Therefore, this explain why 4C piledrivers are being replaced by 2C SR.

When AMD claimed that SR is 2x Jaguar? And what does it mean? Per clock? Per watt? 4x2.0GHz vs. 2Mx4.xGHz?

 

[jdwii]

4 x Jaguar cores = between half and 80% of a Pentium mobile dual core see http://www.anandtech.com/show/6981/the-kabini-deal-can-amd-improve-the-quality-of-mainstream-pcs-with-its-latest-apu/2 I expect SR to be in another league. Jaguar is a competitor to Atom a good one but not good compared to high end desktop CPUs.

Anandtech uses biased benchmarks (Cripple_AMD function) that favour intel chips or uses the ones where AMD chips don't perform so well and ignores the benchmarks where AMD excels. In that review they use pcmark, cinebench, 7-zip...

If you add 33% to the Kabini numbers to simulate the 2Ghz model the results are pretty good. It passes the Pentium in multithreaded benchmarks and still uses less power.

The pentium runs at 2.4GHz and the i5 does up to 2.6GHz. For a clock-clock comparison of raw performance, I would add more than a 33%. Moreover, the effect of the biased benchmarks would be considered as well. It is better to start from fair jaguar benchmarks. Those show that:

4 JG ~ 2 SB + HT
8 JG ~ 4 SB + HT

According to AMD, SR has twice the performance of jaguar; therefore, I think that it is admissible that a 4 SR CPU will be close to an i7-2600k, under fair benchmarks and if everything scales fine.

cowboy44mag
12 July 2013 19:28:27

Do you actually know what you are saying. A 4 core sandy bridge i7 2600k with ht beats amd fastest desktop cpu the fx8350. Now you are trying to say thats what 8 core jaguar cores that are competing against intel atoms or core i3 ulv will be beating a sandy bridge 4c 8t cpu lol. Its a bit like me saying an intel atom 8 core on 22nm will beat the fx8350 when its mles away performance wise.

An i3 3220 will most likely beat the 8 core jaguar easily on most tasks.

In one sense, this confirms my point. Consider a hypothetical FX-4300 downclocked to 1.5 GHz, this would have about the same score than the jaguar. And AMD claims that SR is 2x jaguar. Therefore, this explain why 4C piledrivers are being replaced by 2C SR.

When AMD claimed that SR is 2x Jaguar? And what does it mean? Per clock? Per watt? 4x2.0GHz vs. 2Mx4.xGHz?

If they meant per clock we better all jump board.

 

[jdwii]

I'm just hoping with Steamroller Amd won't be sacrificing Single threaded performance for multithreaded performance. You people do know that even the Phenom has better performance per clock compared to Piledriver in many cases. If steamroller finally puts Amd around even the first gen I7 in performance per clock i'll be happy.

And keep the high clock rates to.

 

[szatkus]

I'm just hoping with Steamroller Amd won't be sacrificing Single threaded performance for multithreaded performance. You people do know that even the Phenom has better performance per clock compared to Piledriver in many cases. If steamroller finally puts Amd around even the first gen I7 in performance per clock i'll be happy.

And keep the high clock rates to.

If that http://www.tomshardware.com/news/AMD-Kaveri-APU-Gaming,22947.html would be true it's more like slighly better Phenom rather than Nehalem per clock.

 

[-Fran-]

I'm just hoping with Steamroller Amd won't be sacrificing Single threaded performance for multithreaded performance. You people do know that even the Phenom has better performance per clock compared to Piledriver in many cases. If steamroller finally puts Amd around even the first gen I7 in performance per clock i'll be happy.

And keep the high clock rates to.

Yes and no... PD has a better showing than Barcelona using newer instructions when you want to compare. It's like the "infamous" SQL benchmark used a couple of posts ago. If you stack up Phenom II against PD in the coming years, we will be indeed surprised by how much margin PD will be ahead.

I agree that at this point and time, PhII has a better showing most of the time, but down the road, when Software (games more specifically) actually updates itself (it will now, since new gen Consoles arrived) we'll see that PD is a better buy hands down. And no, this is not about "threading", but that's also part of the equation here.

Also, when newer games come out, Sandy and maybe Ivy will be finally behind Haswell in every metric. We just need a refresh on the software side.

Cheers!

 

[griptwister]

LOL! Just found this...

http://www.tomshardware.com/forum/id-1722484/tek-syndicate-expert-conjecture-speculation.html

And this I found to be a bit interesting. I'm intrigued by this whole "Cloud Gaming" theme AMD is trying to start. Hope it works out for them.

http://www.ign.com/articles/2013/07/12/ubisoft-ceo-thinks-well-see-another-console-generation?utm_campaign

 

[juanrga]

A 4 core sandy bridge i7 2600k with ht beats amd fastest desktop cpu the fx8350.

This part is also incorrect. Look at Passmark scores: The i7 is behind the FX

http://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-2600K+%40+3.40GHz&id=868

Look to this game

The i7 is again behind the FX. It is possible to find benchmarks where the FX would be a 40 or 50% faster than the i7.

When AMD claimed that SR is 2x Jaguar? And what does it mean? Per clock? Per watt? 4x2.0GHz vs. 2Mx4.xGHz?

From recent interview:

AMD is saying very little about the Steamroller core at this point, except that it offers double the performance of the Jaguar core and will max out with twice the memory capacity, too.

http://www.theregister.co.uk/2013/06/18/amd_opteron_arm_server_chips/

 

[GOM3RPLY3R]

griptwister said:
LOL! Just found this...

http://www.tomshardware.com/forum/id-1722484/tek-syndicate-expert-conjecture-speculation.html

And this I found to be a bit interesting. I'm intrigued by this whole "Cloud Gaming" theme AMD is trying to start. Hope it works out for them.

http://www.ign.com/articles/2013/07/12/ubisoft-ceo-thinks-well-see-another-console-generation?utm_campaign

Thanks for advertising my thread ^_^ lol. I went a little rage on there but held back and posted a "formidable" reply.

Also, that cloud gaming is really a turn for the better. The only thing that still stinks is the CD function. Using games that are downloaded can really stir up some problems with hackers. SONY needs to pump up their security.

I had an idea a while back involving codes and stuff for gaming. I.e Say Jimmy bought game X, and Randy bought Game Y. They love their games, but they love the others games as well. Many of you know, the "Extreme Sharing" that PSN had going until SONY restricted it to two systems.

The idea is that they both can share their games with no problems. So say Jimmy has game X. He will get a code for it every Z months. He has to put that code in order to play. It's like an online pass except for the whole game that expires every Z months. If you want to play offline, you can do so with no problem. The only way to share your game than will be to be online, and using "The Cloud" like Xbox is showing us, you can download your friends game to their system.

Now here's the fun part. Randy now has Jimmy's game X. So now, Jimmy has to use a code that goes into his account once every, lets say one month, to make sure Randy's copy of game X works. Also in turn, every day Randy's system has to use the same code to match up (or a different code that links), to Jimmy's account to authorize it, and for him to play the game. Same thing with offline, you can play it offline.

It's kind of complex, but I can see that it would work out. The code proof system will help with hackers and pirates. Other than that, there are other security measures that may be added. ^_^

 

[szatkus]

So they was talking about Opteron parts. According to this roadmap (http://community.amd.com/servlet/JiveServlet/showImage/38-2545-2550/AMD+Server+Roadmap.jpg) apparently they meant that some 65W Berlin will have 2x performance of X2150 (4 1.9GHz Jaguar cores). Nothing suprising and not too accurate.

 

[GOM3RPLY3R]

http://www.anandtech.com/bench/Product/287?vs=697
i7 2600k beating the fx8350 overall quite easily. Intels main advantage is the significant performance per watt lead they have. This will be shown in performance only when the 8 core haswell cpu gets released. Steamroller should get to 20% within sandy bridge levels at best.

This is where buying an Intel CPU (the right one for the job) counts. The the performance per core per watt ratio, its unbeatable.

 

[The Q6660 Inside]

http://www.anandtech.com/bench/Product/287?vs=697
i7 2600k beating the fx8350 overall quite easily. Intels main advantage is the significant performance per watt lead they have. This will be shown in performance only when the 8 core haswell cpu gets released. Steamroller should get to 20% within sandy bridge levels at best.

This is where buying an Intel CPU (the right one for the job) counts. The the performance per core per watt ratio, its unbeatable.

The issue is, nobody cares about power. Once again, it will take at least six years (On hajifur's "UK power costs" to even get to the 4770K's INITIAL price. Not to mention, the 4770K consumes power too :3 Now hajifur, please leave if you are going to continue with this garbage or we will revert to our secret thread.

 

[GOM3RPLY3R]

http://www.anandtech.com/bench/Product/287?vs=697
i7 2600k beating the fx8350 overall quite easily. Intels main advantage is the significant performance per watt lead they have. This will be shown in performance only when the 8 core haswell cpu gets released. Steamroller should get to 20% within sandy bridge levels at best.

This is where buying an Intel CPU (the right one for the job) counts. The the performance per core per watt ratio, its unbeatable.

The issue is, nobody cares about power. Once again, it will take at least six years (On hajifur's "UK power costs" to even get to the 4770K's INITIAL price. Not to mention, the 4770K consumes power too :3 Now hajifur, please leave if you are going to continue with this garbage or we will revert to our secret thread.

I'm concerned about power? I want to know what PSU I may need. And what about Overclocking? There are many factors relative to power. They may be minor, but they are factors.

Speaking of power, I think it's kind of strange that when I plugged this into a few calculators:
3930k @ 4.5 Ghz 1.25 v-core
Dual SLI GTX 780
and with the 2 HDDs, 1 SSD, and h100i, it all came out to ~ 850 watts.

I was kind of scared as somehow a 3770k @ 4.5, 1.2vcore and Dual SLI 680s gets ~950?

I tried it on a few calculators, they all said the same, I'm kind of scared. Lol

 

[cowboy44mag]

http://www.anandtech.com/bench/Product/287?vs=697
i7 2600k beating the fx8350 overall quite easily. Intels main advantage is the significant performance per watt lead they have. This will be shown in performance only when the 8 core haswell cpu gets released. Steamroller should get to 20% within sandy bridge levels at best. I think the way intel are going they are focused on beating Arm for small devices. They know already they got amd covered for a few years at least.

Man you really do like to troll. By the way when I said 2 core SB with HT is ~comparable to 4 core Jaguar (i3 SB ~ 4 core Jaguar) and 4 core SB - never said i7, they did have 4 core i5 SB did they not? 4 core SB (i5) ~ to 8 core Jaguar. That is assuming mulit-thread benchmark. (by the way in case your confused, which you seem to be a lot, ~ means roughly equal to) I think everyone knows the single thread benchmark would be hands down SB.

Anyway getting back to this trolling remark you really think that Steamroller will only come to within 20% Sandy Bridge levels? I love how "god like" Intel users think they are.... FX 8350 already outperforms SB where it counts multi-thread benchmarks. Others have said it, and they are right when software actually "catches up to" the hardware the machines that can handle multi-threaded tasks the best are going to be the ones that shine. For those who still cling to the idea that single thread performance is the only thing that matters, we already know that Steamroller is going to have much better single core performance than the FX 8350.

Make no mistake Steamroller FX may not be as good as Ivy Bridge, but it will be better than Sandy Bridge in performance per clock. I know, your going to jump on the performance per watt thing, but like I already said where I live I don't have to power my ranch by a little mouse running on a wheel powering a wee little generator.

 

[cowboy44mag]

http://www.anandtech.com/bench/Product/287?vs=697
i7 2600k beating the fx8350 overall quite easily. Intels main advantage is the significant performance per watt lead they have. This will be shown in performance only when the 8 core haswell cpu gets released. Steamroller should get to 20% within sandy bridge levels at best.

This is where buying an Intel CPU (the right one for the job) counts. The the performance per core per watt ratio, its unbeatable.

You know a study that I would love to see performed is who actually uses what they have. What I mean by that is take the guy who lives in the city, never goes into the country, never goes mudding, never goes to the mountains, never goes off road at all but he has a full size (no H2 for this waste of humanity) Hummer jacked up 10" and is usually painted a bright yellow as if that monstrosity isn't attracting enough attention already. Is that vehicle being used for what it was created for? No, because for that guy any car will be able to get him where he is going. All the awesome off road capabilities are totally wasted. 90% of Intel users fall into the same category.

I could easily afford to go online right now and buy the most expensive Intel processor and motherboard combo available, but I would never get my "money's worth" out of it. Heck I'm still running every game I want on ultra settings with my old Phenom II 965 BE, other than gaming, running the internet, a little video encoding and word processing I don't use my computer for too much. Its not running "professional level" programs that would actually take advantage of Intel's performance per clock superiority. Yes there are professionals out there that NEED that additional performance per clock (given the nature of this thread maybe some posters here), but they make up a very small dynamic of overall buyers. I would say that 90% off all people who spent way more than they needed to for their Intel system will never utilize that systems full potential or power... driving a Hummer on city streets.

That is why I'm following Steamroller's release so closely. Steamroller probably won't be as powerful clock per clock as Ivy Bridge or Haswell, but it will sure run anything that I (or 90% of all overall users) will ever need it too. Its multi-core / multi-thread capability may be the best in the industry. I have already got 4 great years out of my Phenom II, and I'm sure I'll get the same enjoyment out of Steamroller, so you guys can't say AMD has no longevity. When you buy something you buy it to do what you need it to do, anything more is a waste of money and product.

 

[juanrga]

So they was talking about Opteron parts. According to this roadmap (http://community.amd.com/servlet/JiveServlet/showImage/38-2545-2550/AMD+Server+Roadmap.jpg) apparently they meant that some 65W Berlin will have 2x performance of X2150 (4 1.9GHz Jaguar cores). Nothing suprising and not too accurate.

The quote is about cores "the Steamroller core [...] offers double the performance of the Jaguar core".

Seattle is a replacement for X2150. Berlin is a replacement for 3300 series. This quote from Feldman is also interesting:

Berlin is cool, and it uses a new Steamroller core from us and delivers tremendous compute and power efficiency

This does not sound uninteresting to me.

http://www.anandtech.com/bench/Product/287?vs=697
i7 2600k beating the fx8350 overall quite easily.

Irrelevant, because Anandtech uses lots of biased benchmarks with the Cripple_AMD function (e.g. Cinebench) or with fake scores (e.g. Sysmark), instead of realistic and honest benchmarks. This is how Sysmark has been cheated to favour Intel chips

Anandtech also selects some benchmarks favouring Intel. If anything is favouring AMD then is eliminated from the list. Look to their Photoshop benchmark now look at this

The FX-8350 is slightly faster than the 3770k; therefore will be faster than the 2600k. Or look to this pair of Anandtech benchmarks

Are this pair of benchmarks added to the list given by the [strike]troll[/strike] hafijur? No, only one of them is added. Now guess which of both benchmarks was added... yes the one where Intel wins, the other was ignored.

Stop from posting nonsense and biased sites. People here is not idiot.

 

[$hawn]

Talking about benchmarks, here's something worth looking at,

http://www.eetimes.com/author.asp?section_id=36&doc_id=1318894
http://www.phonearena.com/news/Intel-Atom-Z2580-AnTuTu-score-drops-20-after-revision-to-site_id45271

 

[juanrga]

Talking about benchmarks, here's something worth looking at,

http://www.eetimes.com/author.asp?section_id=36&doc_id=1318894
http://www.phonearena.com/news/Intel-Atom-Z2580-AnTuTu-score-drops-20-after-revision-to-site_id45271

Yes some days ago Intel was caught trying to cheat another benchmark:

http://www.eetimes.com/author.asp?section_id=36&itc=eetimes_sitedefault&doc_id=1318857

Some additional interesting info here:

http://www.bdti.com/InsideDSP/2013/07/11/JeffBierImpulseResponse

What we found was very interesting: on the Intel Z2580 processor, the compiler removed a key element of the benchmark. Where the benchmark source code calls for a read-modify-write operation, the compiler substituted a write operation alone. It's not yet clear why the compiler omitted some essential steps in the benchmark code.

Once again it is shown that Intel chips are not so good as some make us believe

 

[The Q6660 Inside]

http://www.anandtech.com/bench/Product/287?vs=697
i7 2600k beating the fx8350 overall quite easily. Intels main advantage is the significant performance per watt lead they have. This will be shown in performance only when the 8 core haswell cpu gets released. Steamroller should get to 20% within sandy bridge levels at best.

This is where buying an Intel CPU (the right one for the job) counts. The the performance per core per watt ratio, its unbeatable.

The issue is, nobody cares about power. Once again, it will take at least six years (On hajifur's "UK power costs" to even get to the 4770K's INITIAL price. Not to mention, the 4770K consumes power too :3 Now hajifur, please leave if you are going to continue with this garbage or we will revert to our secret thread.

I'm concerned about power? I want to know what PSU I may need. And what about Overclocking? There are many factors relative to power. They may be minor, but they are factors.

Speaking of power, I think it's kind of strange that when I plugged this into a few calculators:
3930k @ 4.5 Ghz 1.25 v-core
Dual SLI GTX 780
and with the 2 HDDs, 1 SSD, and h100i, it all came out to ~ 850 watts.

I was kind of scared as somehow a 3770k @ 4.5, 1.2vcore and Dual SLI 680s gets ~950?

I tried it on a few calculators, they all said the same, I'm kind of scared. Lol

Power usage can be quite a bit odd, in the end, the GPUs are the real culprit. I must admit, AMD GPUs in CF (not like it works now :lol I managed to help someone upgrade to a 7850 and 8320 on a 550W, it does not get worse from there, only better. Ex:550W+3930K/3770K + Same GPU =8350 + 550W + Same GPU as it is not really worrisome. Heck, I heard a 5870 ran with a i7 920 on a 480W PSU.

 

[The Q6660 Inside]

http://wccftech.com/amd-centurion-fx-9590-5-ghz-processor-previewed-close-core-i74770k/
http://www.3dmark.com/3dm11/6104138

Intels laptop cpu 3dmark11 physics score still beats the the fx9590. Intel may have to release 6 cores soon to the mainstream.

>implying different GPUs
>implying the 9590 or 3940XM was not a victim based CPU.
>implying people with brains will not get the 8350 and clock that to 4.7
>implying mainstream 6-cores from a greedy company
>implying the 9590 is just another Phenom II TWKR scheme
>implying Intel beats AMD in everything beyond the 4770K
>implying Opterons lose to i3s in multithreaded tasks that support all of the Opteron's power
>implying Intel is "WAAAAY superior" than AMD in basically all tasks
>implying performance per watt counts for a desktop system, enjoy your novelty
http://i.imgur.com/AEIqCA6.jpg

 

[MU_Engineer]

I'm just hoping with Steamroller Amd won't be sacrificing Single threaded performance for multithreaded performance. You people do know that even the Phenom has better performance per clock compared to Piledriver in many cases. If steamroller finally puts Amd around even the first gen I7 in performance per clock i'll be happy.

And keep the high clock rates to.

Single-threaded, per-clock comparisons aren't very useful with today's multi-cored CPUs with grossly different architectures and clock speeds between manufacturers and even different generations from the same maker. The performance of a particular chip in a wide variety of programs is really what counts as that's what people see in real-world usage. That means a mix of increasingly multithreaded programs with very different characteristics. It is very difficult to even get an accurate, repeatable single-threaded benchmark today since it takes specialized tools polling CPU registers to even know how fast the CPU is running at a certain period of time. And then as we have learned, first with Intel's HyperThreading and then later with the original Phenom's per-core frequency scaling and Bulldozer's CMT modular architecture, the OS's scheduler has a big hand to play in how well a chip performs with poorly-threaded tasks as well. The single-threaded, "instructions per clock" type of comparisons are really only for academic interest today. The real useful benchmarks are the "let's run a real program in a manner in which actual users might run it and then take measurements." The 1080p+ gaming benchmarks with high detail and a decent GPU are generally representative of this, and unsurprisingly, don't show a huge amount of difference between analogous CPUs of the two makers.

So they was talking about Opteron parts. According to this roadmap (http://community.amd.com/servlet/JiveServlet/showImage/38-2545-2550/AMD+Server+Roadmap.jpg) apparently they meant that some 65W Berlin will have 2x performance of X2150 (4 1.9GHz Jaguar cores). Nothing suprising and not too accurate.

That roadmap says a few things but doesn't make any mention of the performance of the Steamroller APU vs. Jaguar. All it says is that uniprocessor will move from AM3+/the 4M Piledriver die to FM2+/2M Steamroller, Jaguar SoCs will be replaced with ARM units, and that there will be a Piledriver refresh on C32/G34 but no Steamroller.

I am a little curious what the Piledriver refresh yields on C32/G34. They are keeping the same sockets so I doubt there will be any change in I/O, they are not increasing module count, and I also highly doubt there will be a platform refresh since supposedly Piledriver is the last arch supported on C32/G34 before they move to the new "GC36" socket with on-die PCIe and such. My guess is it's a new C stepping with slightly higher clocks and maybe a few small tweaks to things like caches, FPUs, and branch predictors to subtly improve performance. I suppose they could boost the L3 cache size but I haven't heard of the cache size being a significant issue in these chips.