AMD Bulldozer FX-8120 Benchmarks

[BlueBlazer]

In the last few hours, its raining FX benchmarks

Latest find >> FIRST in Southeast ASIA : AMD Bulldozer FX-8120 Benchmarks

Overclocking on air.....

Though the poster seems legit, as usual take it with a grain of salt. :hmm:

 

[BlueBlazer]

Here's another one from VR Zone >> AMD Bulldozer FX-8120 exclusive exposure!. The benchmark scores are lower than the first post (perhaps due different motherboard, RAM or BIOS). It also seems that the Cinebench scores are affected by C-states. :\

 

[nonameo]

That cinebench looks rather disappointing given bulldozer is supposed to excel at multithreaded applications.

 

[Riek]

That cinebench looks rather disappointing given bulldozer is supposed to excel at multithreaded applications.

Simply by looking at the Wprime scores and they say they speak about the same cpu should ring a bell.
10 vs 18.. thats the same cpu being 80% slower?
300 vs 560...

 

[nonameo]

Simply by looking at the Wprime scores and they say they speak about the same cpu should ring a bell.
10 vs 18.. thats the same cpu being 80% slower?
300 vs 560...

Also, looking at the "multiprocessor speedup" suggests that perhaps there is too much resource contingency going on. ~4-5x multiplication on an 8 core?

???

 

[inf64]

Also, looking at the "multiprocessor speedup" suggests that perhaps there is too much resource contingency going on. ~4-5x multiplication on an 8 core?

???

Results are weird indeed (and not because of that speedup since it's inline with design: 4.8x speedup is due to 4 256bit FlexFP units and in single thread mode one core takes the whole 256 FPU for itself).
Results are weird because VRzone posted a news link with supposedly the same CPU on different board that has ~80% lower performance?? Wprime is roughly 80% slower on the same chip that runs on different motherboard... And the "better" results are still worse than 6C Thuban @ 3.2Ghz.
I suspect something is going on with those BIOS versions.

 

[trollolo]

y'all guys just mad that you didn't wait for bulldozer and bought into sandy bridge. who cares if it performs terribly, i'm going to have twice as many cores as you fools

That'll be enough thread crapping.
-ViRGE

 

[BlueBlazer]

Simply by looking at the Wprime scores and they say they speak about the same cpu should ring a bell.
10 vs 18.. thats the same cpu being 80% slower?
300 vs 560...

Yes, I wondered about that also. One is from known hardware guy (from his profile, he sells PC hardware including CPUs), while the other is from "Product Testing" (again looks legit). :hmm: Though different scores, both have something in common and that is the scores are lower than the top Thuban CPU (as rumored). I can only speculate that they have either....

- different motherboard, or different BIOS (immature BIOS?)
- different B2 stepping Zambezi (early B2 and B2F?)
- perhaps different DDR3 memory

Getting confused...

 

[nonameo]

Results are weird indeed (and not because of that speedup since it's inline with design: 4.8x speedup is due to 4 256bit FlexFP units and in single thread mode one core takes the whole 256 FPU for itself).
Results are weird because VRzone posted a news link with supposedly the same CPU on different board that has ~80% lower performance?? Wprime is roughly 80% slower on the same chip that runs on different motherboard... And the "better" results are still worse than 6C Thuban @ 3.2Ghz.
I suspect something is going on with those BIOS versions.

ah, is cinebench FP heavy? I wouldn't know... in that case the scores don't look so bad

 

[inf64]

Chip is the same I think so it's not that. It's the boards.

ah, is cinebench FP heavy? I wouldn't know... in that case the scores don't look so bad

Oh they are bad alright,since it still can't beat X6 @ default (1090T). Something is borked with those systems and I think it's the firmware.

 

[PreferLinux]

Nice max TDP in CPU-z??? Including different between the two!

 

[BlueBlazer]

Nice max TDP in CPU-z??? Including different between the two!

Quote from link......

CPU-Z 1.58 version there were some errors, including the Name, Max TDP (actually 95W), but overall the data are quite accurate, and perhaps some small errors will have to wait until the updated version of the update

 

[LondonBurning]

I'm waiting for official benchmarks. Years in development and slower than an already kind of slow Phenom II?I doubt it...

 

[Nemesis 1]

Results are weird indeed (and not because of that speedup since it's inline with design: 4.8x speedup is due to 4 256bit FlexFP units and in single thread mode one core takes the whole 256 FPU for itself).
Results are weird because VRzone posted a news link with supposedly the same CPU on different board that has ~80% lower performance?? Wprime is roughly 80% slower on the same chip that runs on different motherboard... And the "better" results are still worse than 6C Thuban @ 3.2Ghz.
I suspect something is going on with those BIOS versions.

For a guy that no hardware as you do and you seem to no my local rep. Stop over to the shop . I show you what you know.

 

[inf64]

This message is hidden because Nemesis 1 is on your ignore list.

Sorry can't read your rabble rabble.

If someone is on your ignore list, keep it that way by not rubbing the fact in their face. This is inflammatory behavior and isn't going to be tolerated.
Anandtech Moderator - Keysplayr

Yes, I wondered about that also. One is from known hardware guy (from his profile, he sells PC hardware including CPUs), while the other is from "Product Testing" (again looks legit). :hmm: Though different scores, both have something in common and that is the scores are lower than the top Thuban CPU (as rumored). I can only speculate that they have either....

- different motherboard, or different BIOS (immature BIOS?)
- different B2 stepping Zambezi (early B2 and B2F?)
- perhaps different DDR3 memory

Getting confused...

I think the CPUs are the same.The only variable left is motherboard.

 

[gmaster456]

I'm waiting for official benchmarks. Years in development and slower than an already kind of slow Phenom II?I doubt it...

Same here. I'm kinda skeptical of these supposed "leaked benchmarks" I'm waiting for the actual CPU's to be released before I make my decision.

 

[soccerballtux]

alright so for comparison guys, my ph2 965 gets 10.3-10.7s in wPrime 32million clocked at 4ghz + 2.8ghz CPU-NB (L3cache).

 

[BlueBlazer]

I'm waiting for official benchmarks. Years in development and slower than an already kind of slow Phenom II?I doubt it...

This is one of the rumors for the delays >> Computex: "Bulldozer AMD Phenom slower than X6"....

From conversations we would face several motherboard manufacturers have conducted during the Computex exhibition, we learn that AMD is currently trying with might and the performance of its processors to enhance Dozer before they are officially introduced. The reason for the delay is that recent performance at the moment of a too low level.

From two different sources - which we obviously do not know them - we were told that the current 8-core AMD Bulldozer engineering samples to motherboard makers as they are less performance than the current six-core Phenom II X6 1100T. With such performance, the introduction of the new generation of processors have relatively little effect. With the clock frequency on the sample-we know not, but according XBIT labs , the default clock speed for the 2.5 GHz samples B0 and B1. AMD might try to include the clock frequency of CPUs continue to increase through a B2 revision.

About when we actually expect CPUs now we get different stories. Some sources speak of August, but most of September.

It does seem more or less confirmed. :hmm:

Same here. I'm kinda skeptical of these supposed "leaked benchmarks" I'm waiting for the actual CPU's to be released before I make my decision.

Then you will have to wait until launch (possibly somewhere in mid-October), with a different B2 stepping (sources including Anand says "B2G" stepping).

 

[gmaster456]

This is one of the rumors for the delays >> Computex: "Bulldozer AMD Phenom slower than X6"....It does seem more or less confirmed. :hmm:

Then you will have to wait until launch (possibly somewhere in mid-October), with a different B2 stepping (sources including Anand says "B2G" stepping).

I'm perfectly fine with waiting. After all I have been since May :'(. Regaurdless of the outcome, its better to wait and see what happens. I'm not going to go out and blindly buy an AM3+ board without knowing whats to come for the socket and whether or not my money would be better spent on a Sandy Bridge mobo instead. I want to go with AMD but they really have to deliver. Otherwise I will have no choice but to go with intel.

 

[AdamK47]

I'm regularly disappointed in the benchmarking abilities of those in Asia with pre-released hardware. Is not possible to run something meaningful with the hardware?

 

[Tsavo]

I'm regularly disappointed in the benchmarking abilities of those in Asia with pre-released hardware. Is not possible to run something meaningful with the hardware?

+1^20

 

[dac7nco]

I have doubts of bullshit Asian benchmarks

Fixed.

I have problems believing even AMD would be so foolhardy to spend this long on a a new from-the-ground-up Architecture if is was blown out of the water by an 8c Shanghai (2x K10) system. All of the benches we've seen so far are suspect, and I doubt I'll pay attention anymore until I see Anand post a breakdown.

XBit and a lot of others have broken their credibility by pandering to those of us (me too) who are nearly desperate to see SB-E or Bulldozer anything. I'm really, really hoping this won't be another Phenom launch, but...

Daimon

 

[BlueBlazer]

I'm perfectly fine with waiting. After all I have been since May :'(. Regaurdless of the outcome, its better to wait and see what happens.

That old article (from May) seems to be the most accurate so far (including the B2 stepping and September month mentioned, whichever way to look at it). Too bad, many other rumor and news sites (like Semiaccurate) did not pick this up (probably because it looks too bad to be true).

I'm not going to go out and blindly buy an AM3+ board without knowing whats to come for the socket and whether or not my money would be better spent on a Sandy Bridge mobo instead.

Concerning "blindly buy an AM3+ board without knowing whats to come" >> here's a little humorous post I saw. Cheers.

 

[RussianSensation]

I

That can't possibly be real for a 3.1ghz BD with 4.0ghz Turbo Boost?

Here is my stock E6600....

Overclocking on air.....

As I hypothesized earlier: AMD won't have significant overclocking advantage over SB on 32nm. 4.7ghz @ 1.40V is nothing special. Not sure how AMD expects to compete if they have not made progress in IPC and their overclocking on 32nm is not any better than SB. :'(

Here's another one from VR Zone >> AMD Bulldozer FX-8120 exclusive exposure!. The benchmark scores are lower than the first post (perhaps due different motherboard, RAM or BIOS). It also seems that the Cinebench scores are affected by C-states. :\

If those benches are real (i.e., FX-8120 3.1ghz/4ghz Turbo) is barely faster than a stock i7 860 2.8ghz (2.93ghz Turbo with 4 cores), then BD's IPC is worse than Phenom II. That would be a failure of MASSIVE proportions.

 

[RussianSensation]

This can't be real, no way!

An 8 core 3.1ghz chip can't beat a 4 core 3.3ghz chip?

This would literally mean that 4 core SB is faster than an 8 core BD. Even in single threaded cinebench there is a 60% advantage in favor of the 2500k.

Worse IPC than Phenom II?