Bulldozer Review! Legit?

[Caza]

You're probably right, but a Deneb core couldnt support whats coming in the future or support what BD does, i still believe that the current implementation in zambezi is borked somewhere.

I'm not sure if borked is the right word but it definitely has a performance issue. An AMD engineer I spoke with said it has already been fixed and in the working Trinity APU engineering samples. It just wasn't corrected in time for the Zambezi launch, however it may follow shortly with the next production run when higher clock speed models are released. Bulldozer enhanced core I've seen it referred to as, will offer a few more tweaks before they move on to Piledriver.

 

[Despoiler]

I'm not sure if borked is the right word but it definitely has a performance issue. An AMD engineer I spoke with said it has already been fixed and in the working Trinity APU engineering samples. It just wasn't corrected in time for the Zambezi launch, however it may follow shortly with the next production run when higher clock speed models are released. Bulldozer enhanced core I've seen it referred to as, will offer a few more tweaks before they move on to Piledriver.

This is like the last thing I wanted to hear; that we have to wait for the 8170 for the performance to be what it should.

 

[grimpr]

This claims to but I don't have any AMD CPUs to test it with.

What a great find! I tested it on Witcher 2 and it found it as an Intel Compiler program! Well, off for more tests...

 

[Imouto]

IMHO,
taking only a single app/bench(Cinebench 11.5) and making a general performance conclusion, will only make you look silly after the official FX reviews.

QFT

And not only him, everyone making these kind of statements 4 days away of launch date. I would extend it to any kind of statement regarding performance.

 

[NostaSeronx]

QFT
I would extend it to any kind of statement regarding performance.

Then point out R11.5 applies to only R11.5

Cinema 4D R12 and R13 have been out for awhile

R12:

[LIST]
[*]Dynamics has been improved (complex physicengine)
[*]Renderer works better / faster.
[*]Menu has been cleaned up.
[*]Many minor improvements like the Shaders and Materials.
[*]Linear workflow introduced with correct display in OGL.
[*]Double precision calculation introduced and true units.
[*]Support for IES lights and physically correct lighting.
[*]Integration of [URL="http://en.wikipedia.org/wiki/Python_%28programming_language%29"]Python[/URL]
[*]Support of [URL="http://en.wikipedia.org/wiki/OpenGL_3"]OpenGL 3[/URL]
[*]Support for PPC CPUs (G5) dropped.
[/LIST]

R13:

[LIST]
[*]New Physical Render engine (More photorealistic: Physical Camera  with ISO, f-stop and shutter speed or angle, true depth of field, true  motion blurring, lens distortion, vignetting, and chromatic aberration.)
[*]Stereoscopy. New Stereoscopic Camera and Rendering with real-time  display. Supports Anaglyph, Side-by-Side, Interlaced, or Active Shutter  mode for 3D Monitors, projectors and glasses.
[*]New Character tools, allowing the user to set up complex rigs  quickly and easily. CMotion Walk Cycle editor; Character Components for  TDs to define rig hierarchies, and new Muscle Object & MSkin  Deformer.
[*]Animation refinements. Rotation Order and Gimbaling Rotation Axis to  detect and correct gimbal lock. Better timeline markers. Finely tuned  F-Curves. Colored animation paths.
[*]Optimized Modeling workflow. (Enhanced cloning and extrusions;  Advanced selection tools; new optimized HyperNURBS weighting algorithm  for cleaner topology and sharper but smooth creases; Axis toggle; Auto  Tweaking; Easy, intuitive Snapping; Variable Speed Scaling; Better  ClothNURBS Extrusion)
[*]New, accurate Sub-Surface Scattering shader (SSS). Terrain Mask shader. Brick Shader enhancements. Enhanced MoGraph Muti-Shader.
[*]Improved Interface and Workflow. Enhanced Picture Viewer. All-new External Referencing System for XRefs.
[*]New Import/Export features (After Effects Integration; FBX  Import/Export supports 2010, 2011, and 2012; Multi-Channel OpenEXR;  Enhanced Sound Format support including AIF, MP3, and AAC; Collada 1.5;  BodyPaint 3D support for Maya UVs)
[/LIST]

 

[jones377]

What a great find! I tested it on Witcher 2 and it found it as an Intel Compiler program! Well, off for more tests...

Maybe this deserves a thread of it's own? people could post before and after scores with AMD CPUs....

 

[Black96ws6]

This claims to but I don't have any AMD CPUs to test it with.

Is this a legitimate program?

My 1090T@3.8Ghz gets 6.74 in CB 11.5 64bit. It would be interesting to run this and see the after result (assuming they use an Intel compiler).

I have a small screenshot of my current results, can you attach images here other than using a URL? I'd like to attach it...

 

[Black96ws6]

Also, here are some more quotes from legitimate sites and reviewers who actually have the product in hand:

The issue with this press kit is the lack of support : CPU-Z is being updated as we speak, no true driver package, etc....eg with the SB launch Intel gave us a complete package of software to read out temps, monitor turbo ( we have overdrive for this ) a complete driver package, working CPU-Z, AIDA,... Now we just have PPTs, 2 biosses and Overdrive... and that's it... new bios we have via Asus contact is far better then the biosses given by AMD. So now I have to retest lol...

what bios are you using? Asus Crosshair V Formula tested 0083 0813 and 9905 bios now (last one gives best performance and opens ram dividers to 2400)

no biggie to boot with 2600NB, 2800 boots only with extreme voltages (board even sets 1.45 on AUTO ) While for 2600 I don't need to adjust voltages at all. 3000 and higher is no post... Thomas has similar issues.

And Turbo doesn't seem to work with 9905

I have 1.58.6 CPU-Z, I heard .7 version is released soon

Overal 9905 is faster in almost every app, New ram dividers work top notch, love that 2400mhz

 

[MikeyMouse]

...

 

[aigomorla]

Specs please, why 3 loops?

as i said ART.

http://forums.anandtech.com/showthread.php?p=31538648#post31538648

is it needed?? no...
Why?

Because watercooling is fun... and its an art when done correctly, as well as a science.

 

[grimpr]

Is this a legitimate program?

My 1090T@3.8Ghz gets 6.74 in CB 11.5 64bit. It would be interesting to run this and see the after result (assuming they use an Intel compiler).

I have a small screenshot of my current results, can you attach images here other than using a URL? I'd like to attach it...

Run Cinebench 10, include the programs librarys (.cdl, .cdl64) too since everything is compiled with ICC, patch them all with the Intel Compiler Patcher, rerun benchmarks, witness pure software magic from Intel.

CB11.5 propably uses more modern extensions that only BD supports. We will see soon enough.

 

[grimpr]

Maybe this deserves a thread of it's own? people could post before and after scores with AMD CPUs....

Sure does, simply because it works and the scan feature is fantastic.

 

[Manoa]

what does Intel Compiler Patcher do ? disable the use of the instructions ?
because I know of an another related program which does something different: it replaces the string in the program "GenuineIntel" with "AuthenticAMD" and with that enables the use of the intel instructions on AMD CPU's. seems a shame to disable the optimizations altogether, problem is that as far as I know this software only runs on linux, the file is called patch-AuthenticAMD.tar.gz but I can't find it for download anymore, the site of the author Ricardo Catalinas Jiménez is down and there does not seem to be a copy of the program available, I could upload it myself, but that is good thinking grimpr I have R10 im going to run some tests with that program myself see how it affects my cpu. the good news is that the Intel Compiler Patcher didn't find any of the standard benchmarks to be Intel applications (prime 95, superPi, sciencemark 2, rmma 3.8)

 

[soccerballtux]

what does Intel Compiler Patcher do ? disable the use of the instructions ?
because I know of an another related program which does something different: it replaces the string in the program "GenuineIntel" with "AuthenticAMD" and with that enables the use of the intel instructions on AMD CPU's. seems a shame to disable the optimizations altogether, problem is that as far as I know this software only runs on linux, the file is called patch-AuthenticAMD.tar.gz but I can't find it for download anymore, the site of the author Ricardo Catalinas Jiménez is down and there does not seem to be a copy of the program available, I could upload it myself, but that is good thinking grimpr I have R10 im going to run some tests with that program myself see how it affects my cpu. the good news is that the Intel Compiler Patcher didn't find any of the standard benchmarks to be Intel applications (prime 95, superPi, sciencemark 2, rmma 3.8)

thought that wasn't possible without virtualization

 

[grimpr]

I'm not sure if borked is the right word but it definitely has a performance issue. An AMD engineer I spoke with said it has already been fixed and in the working Trinity APU engineering samples. It just wasn't corrected in time for the Zambezi launch, however it may follow shortly with the next production run when higher clock speed models are released. Bulldozer enhanced core I've seen it referred to as, will offer a few more tweaks before they move on to Piledriver.

Nailed!

First Look: AMD Trinity APU, Linux Already Runs Well

http://www.phoronix.com/scan.php?page=article&item=amd_trinity_early&num=1

"The Trinity results I've seen under Linux have been impressive, but tough to compare to my current Llano hardware due to other hardware differences. The Catalyst graphics support seems a bit premature, but we're still months away from the actual Trinity launch. The compute performance is at least in order."

"While Trinity may be a very nice step up from Llano and other current-generation hardware, for those wanting top-notch performance, the AMD Eng Sample 1S192145TGG44_3/19/2_2/16 is a beautiful Bulldozer server chip. AMD Eng Sample 1S192145TGG44_3/19/2_2/16 is a 16-core Bulldozer that doesn't appear to be yet on the market. Two of these beauties can be packed into one system and run well under Ubuntu 11.10."

http://openbenchmarking.org/ is where the real action is.

 

[RussianSensation]

IMHO,
taking only a single app/bench(Cinebench 11.5) and making a general performance conclusion, will only make you look silly after the official FX reviews.

I never said that Cinebench will be 100% representative of the final performance. However, it is one data point, along with wPrime and SuperPi and Fritz benchmarks. You add all of those data points together, and unless those benchmarks are actually fake, they are not showing BD as anything extraordinary. Of course if those benchmarks are fake, everything is out the window. However, if they are true, that starts to paint a different picture.

Of course we'll find out in 4 days

 

[sequoia464]

I'm not sure if borked is the right word but it definitely has a performance issue. An AMD engineer I spoke with said it has already been fixed and in the working Trinity APU engineering samples. It just wasn't corrected in time for the Zambezi launch, however it may follow shortly with the next production run when higher clock speed models are released. Bulldozer enhanced core I've seen it referred to as, will offer a few more tweaks before they move on to Piledriver.

Nailed!

First Look: AMD Trinity APU, Linux Already Runs Well

http://www.phoronix.com/scan.php?page=article&item=amd_trinity_early&num=1

"The Trinity results I've seen under Linux have been impressive, but tough to compare to my current Llano hardware due to other hardware differences. The Catalyst graphics support seems a bit premature, but we're still months away from the actual Trinity launch. The compute performance is at least in order."

"While Trinity may be a very nice step up from Llano and other current-generation hardware, for those wanting top-notch performance, the AMD Eng Sample 1S192145TGG44_3/19/2_2/16 is a beautiful Bulldozer server chip. AMD Eng Sample 1S192145TGG44_3/19/2_2/16 is a 16-core Bulldozer that doesn't appear to be yet on the market. Two of these beauties can be packed into one system and run well under Ubuntu 11.10."

http://openbenchmarking.org/ is where the real action is.

If this is accurate, and we will be seeing a new stepping down the road to address the 'performance issue', any idea what sort of time frame that might be?

 

[grimpr]

If this is accurate, and we will be seeing a new stepping down the road to address the 'performance issue', any idea what sort of time frame that might be?

I dont know jack since this is all wild speculation coming from forums, chats, rumours etc, but if i would make a guess, it would be Q1 2012. Current rumours are that with the new bioses around the performance is upped.

 

[NostaSeronx]

I dont know jack since this is all wild speculation coming from forums, chats, rumours etc, but if i would make a guess, it would be Q1 2012. Current rumours are that with the new bioses around the performance is upped.

Seeming as the CPUs that are coming out are made in the 36th week of 2011

September 4/5 -> 10/11
and the CPUs are getting in stock October 16th

I give or take 6 weeks from now these fixed versions will be released as B3s with the C0s coming the next year

 

[sequoia464]

I dont know jack since this is all wild speculation coming from forums, chats, rumours etc, but if i would make a guess, it would be Q1 2012. Current rumours are that with the new bioses around the performance is upped.

Thanks, I really, really, really don't know jack - a guess is all I was looking for, as I don't have a clue as to the time involved in any changes to the chip and then production afterwards for a revised stepping.

@Seronx "I give or take 6 weeks from now these fixed versions will be released as B3s with the C0s coming the next year"

Forgive my ignorance, by C0's, do you mean a major revision, not just a new stepping?

 

[NostaSeronx]

Thanks, I really, really, really don't know jack - a guess is all I was looking for, as I don't have a clue as to the time involved in any changes to the chip and then production afterwards for a revised stepping.

@Seronx "I give or take 6 weeks from now these fixed versions will be released as B3s with the C0s coming the next year"

Forgive my ignorance, by C0's, do you mean a major revision, not just a new stepping?

I am only using what happened to Phenom -> Phenom II X4 -> II X6

Phenom B2 -> B3 -> Phenom II C2 -> C3 -> Phenom II X6 E0

Trinity and Vishera ES being C0s and C1s

FX B2 -> B3 -> FX Next C2 -> C3 -> FX Next+ E0

Certain places are already seeing some hints that C0 and C1 are already faster than B2 and B3 and Llano(which if it had AVX/SSE5 it would beat everything on the market but would you really want a 250 TDP cpu?)

 

[Ancalagon44]

I'm not sure if borked is the right word but it definitely has a performance issue. An AMD engineer I spoke with said it has already been fixed and in the working Trinity APU engineering samples. It just wasn't corrected in time for the Zambezi launch, however it may follow shortly with the next production run when higher clock speed models are released. Bulldozer enhanced core I've seen it referred to as, will offer a few more tweaks before they move on to Piledriver.

If this is true I will laugh at them. They are now many months late - heck to be honest they could have used this new architecture years ago. And now even with being months late, the chip is borked? And you want to sell this to people? Reminds me of Phenom I, which launched underperforming and underclocking.

Funny how the graphics division does so much better - they actually release good performing products on time and on budget.

 

[NostaSeronx]

If this is true I will laugh at them. They are now many months late - heck to be honest they could have used this new architecture years ago. And now even with being months late, the chip is borked? And you want to sell this to people? Reminds me of Phenom I, which launched underperforming and underclocking.

I think it would be easier to point it's more like an Engineer Sample of Fermi

B2 Zambezi = GF090
B3 Zambezi = GF100?
C2 Vishera = GF110
C3 Vishera = GF110(again)
E0 Steamroller = GK100(hopefully it will have a good launch)((or Fermi refresh @ 28nm what are those call GF120?))

 

[aigomorla]

notice how the TROLLS are no longer on the forum?
They have nothing to hide behind now...



So u guys still think i was lying all this time?


<--- long live the X6! <--- im dead serious about this statement.

 

[Nemesis 1]

notice how the TROLLS are no longer on the forum?
They have nothing to hide behind now...



So u guys still think i was lying all this time?


<--- long live the X6! <--- im dead serious about this statement.

Really / What trolls you speaking about . The ones who told ya in Dec . BDs true speed . Or the ones who make believe . I haven't seen them go away . I see trinity name being thrown around . Again by people who have zero access to the cpus . This thing is ugly and when IVY bridge is released it gets uglier still .