VideoHelp Forum
+ Reply to Thread
Results 1 to 13 of 13
Thread
  1. Banned
    Join Date
    Nov 2005
    Location
    United States
    Search Comp PM
    well a Turkish website claims they got their hands on an engineering sample 8 core bulldozer, running at 3.2ghz with a max TDP of 186W (ouch), the benchmarks that concern us the most are the x264 and cinebench results:

    http://www.donanimhaber.com/islemci/galerileri/AMDnin-8-cekirdekli-Bulldozer-FX-islemc...onuclari_7.htm

    if this benchmark result is accurate the 45.5 fps second pass average compares favorably against the 36.3 result for a 2600k and the 48.8 result for a 980x:

    http://www.anandtech.com/show/4083/the-sandy-bridge-review-intel-core-i7-2600k-i5-2500...2100-tested/16

    as a way of sanity checking the results, we can compare the multi-threaded cinebench 10 results for the supposed ES bulldozer (24434 CB) against the results for the i7 2600k (22875) and the 980x (27056).

    http://www.maximumpc.com/article/features/sandy_bridge_washes_ashore

    looking at the fritz chess benchmarks, a 2600k is capable of 13017 kilo nodes, this bulldozer sample is capable of 14197, to me the numbers posted look consistent, if the benchmarks have been faked then these guys did their homework; i personally believe that these numbers are reasonably indicative of what bulldozer will bring to the table once released.

    depending on how amd prices these cpu's, and on how high they are capable of turbo'ing, amd may once again find itself in a position to actually be in the black with their cpu division.
    Quote Quote  
  2. I hope it's true, we need some competition for Intel. Maybe they will push back up Ivy Bridge and SB-E
    Quote Quote  
  3. Before I started my second PC running with core i7, I was pretty happy with discarded Athlon XP 3200+ for Graphics.
    Quote Quote  
  4. aBigMeanie aedipuss's Avatar
    Join Date
    Oct 2005
    Location
    666th portal
    Search Comp PM
    right. some turkish no name, no one has ever heard of website in turkish only scoops ever real tech site. i call bullshit.
    --
    "a lot of people are better dead" - prisoner KSC2-303
    Quote Quote  
  5. Banned
    Join Date
    Nov 2005
    Location
    United States
    Search Comp PM
    Originally Posted by aedipuss View Post
    right. some turkish no name, no one has ever heard of website in turkish only scoops ever real tech site. i call bullshit.
    actually they're not the first ones to have posted supposed benchmarks from ES bulldozer cpu's, there's a chinese site and a few other foreign sites. the only reason you have never heard of them is because obviously you only frequent tech sites that post content in english or languages you understand.

    as i said, if these guys faked all these benchmarks then they made sure to do their homework and make the results consistent so that they fall between a 2600k and a 980x.

    personally, i think it's probably very indicative of what we can expect come September or whenever bulldozer finally arrives.
    Quote Quote  
  6. Originally Posted by deadrats View Post
    if this benchmark result is accurate the 45.5 fps second pass average compares favorably against the 36.3 result for a 2600k and the 48.8 result for a 980x
    Are you sure they were running the same settings? As you well know the settings used have a huge impact on the encoding rate.

    Originally Posted by poisondeathray View Post
    I hope it's true, we need some competition for Intel.
    That's for sure.

    And when the part is release it will not be running at 3.2 GHz with a 186 watt TDP. More likely something like 2.4 GHz and 100 watt TDP. So even if these benchmarks are accurate the delivered parts won't be as fast. But the benchmarks could be indicative at what overclockers can get.
    Last edited by jagabo; 12th Jul 2011 at 08:19.
    Quote Quote  
  7. Banned
    Join Date
    Nov 2005
    Location
    United States
    Search Comp PM
    Originally Posted by jagabo View Post
    Are you sure they were running the same settings? As you well know the settings used have a huge impact on the encoding rate.
    it looks to me like they used the standardized x264 benchmark, now of course if they really wanted to cheat they could edit the script that runs the benchmark and we would never know, but again, based on the fact that the 2600k is a 4 core 8 thread part and that the 980x is a 6 core 12 thread part and the tested bulldozer sample is a 4 module 8 integer core part, it seems like it passes the "logically consistent" test.

    Originally Posted by jagabo View Post
    And when the part is release it will not be running at 3.2 GHz with a 186 watt TDP. More likely something like 2.4 GHz and 100 watt TDP. So even if these benchmarks are accurate the delivered parts won't be as fast. But the benchmarks could be indicative at what overclockers can get.
    amd has said that the top of the line part (which this sample is supposed to be) will be a 125w part and all the other bulldozer cpu's will be 95w parts.

    leaks from amd indicate that the postponement of bulldozer was so they could spin some better steppings and get faster clock speeds; i've seen reports that claim amd is aiming for a 4ghz clock speed for it's top of the line part (don't know if that's a stock clock speed or a turbo speed, the reports made it sound like a stock speed target).

    i know lot's of speculation, but a 4ghz part, with a 125w TDP and if these benchmarks are accurate, could mean that amd may just pull another athlon 64.
    Quote Quote  
  8. Originally Posted by deadrats View Post
    amd has said that the top of the line part (which this sample is supposed to be) will be a 125w part
    Scaling from the 3.2 GHz 186 watt part, a 125 watt part will probably be around 2.6 GHz. From the way current Intel chips overclock it looks to me like they could release faster parts any time they wanted. And isn't the next generation of Sandy Bridge due soon too? Intel's in no danger of losing the throne.
    Last edited by jagabo; 12th Jul 2011 at 23:15.
    Quote Quote  
  9. ArsTechnica got hold of one:
    http://arstechnica.com/gadgets/news/2011/07/bulldozer-prototype-suggests-amd-shooting-...erformance.ars

    a FX-8130P, has a base clockspeed of 3.2GHz. With all four Bulldozer "1.5 core" modules running, the processor can "turbo boost" its speed up to 3.6GHz. When only half of its modules' hardware is active, however, it can crank the speed up to 4.2GHz.Stuffed in an AM3 socket-equipped motherboard and running alongside an NVIDA GTX 580 GPU, the Bulldozer-based processor turned in a respectable 3DMark 11 score of 6265. Its physics score, which is largely CPU dependent, is 7487. That puts its performance between Intel's latest quad-core, Sandy Bridge-based Core i5-2500K (6667) and Core i7-2600K (8152), according to recent benchmarks run by Tom's Hardware.
    The FX-8130P's PCMark 7 score, on the other hand, is just 3045. That doesn't even beat AMD's own A8-3850 clocked at 2.9GHz (3575), let alone Intel's Sandy Bridge parts.
    Quote Quote  
  10. aBigMeanie aedipuss's Avatar
    Join Date
    Oct 2005
    Location
    666th portal
    Search Comp PM
    smells like there was some turkish fudge cooked up.
    --
    "a lot of people are better dead" - prisoner KSC2-303
    Quote Quote  
  11. Banned
    Join Date
    Nov 2005
    Location
    United States
    Search Comp PM
    @jagabo and aedipuss

    reread the article, arstechnica did not get get their hands on a test sample, they are reporting on that turkish sites tests and in fact linked to them.

    furthermore, the quote talks about the pcmark 7 scores of the engineering sample, but said scores, with screenshot, was provided as one of the benchmarks the turkish site performed:

    http://www.donanimhaber.com/islemci/galerileri/AMDnin-8-cekirdekli-Bulldozer-FX-islemc...onuclari_4.htm

    look at the screenshots at the bottom of the article, the pcmark 7 test was performed by the turkish site; they didn't lie or try and cover anything up, they tested with 7 or 8 benchmarks and reported all the results and gave visual evidence.

    we need to remember that this is an engineering sample, normally used for validation purposes, we have no way of knowing the state of it's microcode, how fast the hypertransport links were running; there are reports out that indicate that "bulldozer" will have very fast L1 caches but mediocre L2 and L3 caches, we don't know if that will change with final silicon nor if their's something inherent to pcmark 7's code that casts "bulldozer" in a poor light.

    now i'm not saying that "bulldozer" will walk all over SB, but at least in the top end i think it's realistic to expect a 4 module 8 integer core cpu with a 256 fpu to compete favorably with a 4 core 8 thread processor (2600k), on the mid end ($150-$250) i expect AMD to get slaughtered.

    of course Ivy Bridge will definitely lay a smack down, but AMD has said they will be releasing a 5 module 10 integer core desktop cpu sometime early next year and they have said that "bulldozer 2", due out next july or so will be a "bulldozer" apu, meaning integrated DX11 graphics.

    of course, everyone knows my feelings already; i expect that once someone codes an app that fully exploits Quick Sync software based encoding will be a thing of the past and once that killer app is removed from the desktop as a driving force for upgrade among enthusiast then the only thing left is gaming and quite frankly even a i5 2400 coupled with a high end video card is sufficient for playing games.
    Quote Quote  
  12. Originally Posted by deadrats View Post
    reread the article, arstechnica did not get get their hands on a test sample, they are reporting on that turkish sites tests and in fact linked to them.
    I see. The link didn't work earlier when I tried it. So it wasn't clear who/what Donanim Haber was.
    Quote Quote  
  13. The current SB is a mainstream CPU, not high end part. I don't think Intel is worried; when the early benchmarks came out on the Taiwanese sites a few weeks ago , Intel actually pushed back the SB-E release into next year. They have working 6 and 8 core early stepping engineering samples now, I'm sure they can release it within this year if they felt threatened.

    BTW anything posted by "coolaler" or "JCornell" is legit. They are the ones that post early benchmarks and screenshots of ES chips from Taiwan months in advance of retail releases, and have been doing so for years
    http://forum.coolaler.com/showthread.php?t=268982
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!