Is the ASUS crosshair V+580 giving biased results for the Bulldozer?

[mailpranshu]

Hi Guys,

I have stumbled upon something rather strange. Like any other builder I had been waiting for the FX with bated breath. It has been a disappointment and all sites across the net have thrashed it. Let me summarize some of the obvious pitfalls detailed on Toms, Anands, X bit etc.

1. Performs lower than the 2500k.
2. Power consumption is high.
3. Single core performance is dismal. Worse than even the 1100T in some cases!!!!
4. Gaming performance is downright bad.

I am from an engineering background(electronics) and have an additional finance background as well(MBA finance). I know how companies work and the pricing made no sense to me in any scenario. Let me summarize again.

1. Why Had AMD prices something so high when it knows it will not perform from day 1?
2. Whats the point of making a sure shot loss from day 1?
3. Why is the gaming performance so dismal? Wasn't ultra high resolution gaming its USP?
4. Where is the overclocking? 4.5ish GHz???? That's it???!!!!!!!!
5. Was there any testing done? Even a simple die shrink would have improved heating and efficiency.

So I wanted to have a further look. This was a mystery that needed looking into. I also noticed something else. All the testing on all the sites has been done on this kit. Link below.

http://morethanchessagame.forumotion.com/t1987-amd-fx-bulldozer-press-kit-pictured-stuffed-with-fx-8150-cpu-asus-crosshair-v-formula-and-scorpius-platform-badge

Now, all big sites have been given this kit including Toms. It makes sense because I am guessing Asus has an agreement with AMD to make its mother boards the "official" ones for testing. Also all these sites have used a 580 GTX quoting minimal graphics bottleneck. I agree its a sound rationale.

I confirmed at some other sites as well. The Asus + 580GTX has been used pretty much everywhere.

Its then that I went over to hardwareheaven and had a look. They have stayed away from the press kit and used an Asrock board. Here is the review link.

http://www.hardwareheaven.com/reviews/1285/pg1/amd-fx-8150-black-edition-8-core-processor-vs-core-i7-2600k-review-introduction.html

Now, Toms has come up with results exactly the same as all other sites that used the press kit. Lets have a look at some big changes on the benches.

Also, all sites except hardwareheaven has used a 580Gtx. Lets have a look at some gaming benches. That's what matters to us right?

3DMark 11

http://www.hardwareheaven.com/reviews/1285/pg9/amd-fx-8150-black-edition-8-core-processor-vs-core-i7-2600k-review-3dmark-11.html

Win

http://www.tomshardware.com/reviews/fx-8150-zambezi-bulldozer-990fx,3043-13.html

Massive loss!!!!!

F1 2011


http://www.hardwareheaven.com/reviews/1285/pg11/amd-fx-8150-black-edition-8-core-processor-vs-core-i7-2600k-review-f1-2011.html

Win

http://www.tomshardware.com/reviews/fx-8150-zambezi-bulldozer-990fx,3043-19.html

Massive loss!!!!!

Also when I go through the whole hardware heaven review it gives it a good rating placing it close to the 2600K and above the 2500K. Which in turn is in line with their strategy. It does lose in some benches but is still very very close to the 2500K.

This is extremely strange. Please go through the site yourself.

So to conclude:
1. Can Toms and others please do a test with other 990 boards?
2. Can we please have a comparison with more AMD and Nvidia cards?
3. Who do we trust? Isn't a biased review completely anti against AMD from day 1?
4. Can some one else independently run benches with the 2500K, 2600K and 8150 using different cards and mother boards?
5. Is there a Nvidia Intel nexus to make sure 8150 looks bad? Was the BFBC3 patch doctored for more than just RAGE???( remember Hawx 2 tessellation)

Please post links here. And don't post links for other sites as there is a strong anti AMD bias/ vibe. I think we users need to take things in our own hands and check this out. Can some one use their own personal hardware and run benches?

I am not an AMD fan or a conspiracy nut. I use both builds and use Nvidia as well. So lets keep this discussion open and friendly. No trolling please.

And can the seniors moderators and testers at Toms please step in and explain? Can a new test be done?

 

[lp231]

Just ran Cinebench on my aging system and this is what I found out.
* With all cores enabled, using affinity to assign which cores to run the benchmark, Cinebench will still use all 4 of them, but
in task manager 2 of the disabled cores are idle.
* Setting the Windows to run at 2 cores only from boot file, Cinebench will use only 2 cores.
Time to complete the test can be noticed, but results between the 2 settings aren't that much of a difference.

All stock settings, Win7 Ulti x64/SP1

 

[statikregimen]

Interesting findings, tho within error tolerance. You should take the averages of several runs to see if its consistent. Either way, the effect is even more exaggerated on Bulldozer, in many lightly threaded workloads, because of the module layout, where every 2 cores share certain resources.

Perhaps the most significant example of this, in my personal tests so far, was in the old (but good) game Half-Life 2: Deathmatch...I started a local server, and loaded a random map. Standing in one place, not moving my view at all, I would alt-tab to windows and change the affinity in task manager. The game seems to be optimized for 4 cores:

8 cores, no affinity: 505fps avg
4 cores in 2 modules: 460fps avg
4 cores in 4 modules: 520fps avg

Pretty cool

EDIT: btw, what CPU do you have?

 

[lp231]

I ran the test 3 times each and the scores are consistent like 1.57, 1.59, 1.60
I just posted the results of the first run.
CPU is a ancient s775 Xeon (quad core) @ 2.66GHz
Click "Member configuration" for detail specs

2C/2T average: 1.58 (1.57, 1.59, 1.60)
4C/2T average: 1.59 (1.59, 1.60, 1.60)
4C/4T average: 3.11 (3.11, 3.08, 3.16)

 

[FALC0N]

Interesting findings, tho within error tolerance.

Its just disabling 2 cores on his Xenon. Its not a hyperthreaded part.

Notice it says:

2 cores and 2 threads
4 cores and 2 threads
4 cores and 4 threads

the right column is the active cores. So it makes perfect sense that the first two would produce the same results.

 

[megagabobe1]

a guy that makes good video reviews on youtube called motherboards.org said that 95% of the reviews for the fx series are fake

 

[BaronMatrix]

a guy that makes good video reviews on youtube called motherboards.org said that 95% of the reviews for the fx series are fake

Depends on what he means by "fake."

 

[jimmysmitty]

a guy that makes good video reviews on youtube called motherboards.org said that 95% of the reviews for the fx series are fake

Its a conspiracy!!!! Intel sent out samples before AMD wanted them too and fixed the reviews!!!!!

Love the conspiracy stuff.

 

[megagabobe1]

im not really sure haha, im just telling you what i read so that you know, gathering up opinions and stuff haha

 

[megagabobe1]

this is what he said "Almost all reviews you have seen were fake and based off unavailable CPUS that were recalled. I bought mine from NEWEGG!!!", answering to this : Here are some facts about 95% of the benchmarks for Bulldozer:

1. The CPU used is the FX-8150.

2. They used ASUS Motherboards.

3. The processor they used was an engineering sample.

 

[statikregimen]

Can you provide a link? If its the same guy I'm thinking of, he's struck me as fairly credible and I don't think he'd really say anything like that, unless there was a really good reason.

No matter, though - whether or not they used an engineering sample, we can clearly see from my own benchmarks, that even at 4.5ghz, there's not a leg to stand on - the CPU's IPC/per-core performance is pretty bad and clock for clock, about 5-7% lower than the Phenom II, when it should be 5-10% faster than anything ever made before it.

Believe me, I was really hoping to find some magic pixie dust hidden in the tin, BIOS or some magic OS setting I could change to redeem it, but so far, the best you can do is OC the crap out of it.

However, I will admit, at that point, the performance is pretty good. It does scale decently, if you can keep it cool and don't mind the ridiculous power consumption. I'm happy with the final result, overall.

 

[megagabobe1]

here is the link for the video, read the comments, i just found interesting what he said, so im sharing it to you know guys

http://www.youtube.com/watch?v=pxqO8LoHxT4&feature=feedu

 

[yomamafor1]

Soooo......AMD sent out unfinished (alleged) CPU in their press kits, designed to portrayed themselves in the best light possible, to reviewers.

Ehhhh.... nope, not buying that, at all.

 

[statikregimen]

Ah, yeah, I just watched that video a few minutes ago, when I saw your post here. Now I see what he said in the comments. Interesting, but fake? I'm inclined to believe not...Again: I think my benchmarks speak for themselves. My analysis, at this point is that the motherboard isn't the problem....the CPU just majorly slacks in single-core performance.

Oh, and btw guys: I've managed to squeeze out another 146mhz for 202x23 @ 1.48v for 4.646ghz total...I've tried to get 4.7ghz out of it, and it just wont have it without massive voltage increases beyond what I'm willing to run on air 24/7.

 

[megagabobe1]

yeah, i dont know what to believe, im more inclined to your personal results statikregimen, and to see the review from the guy i showed u when he uploads it

 

[Onus]

Yeah statik, I'm really impressed by the detailed testing you've been doing. The fact that it hasn't really changed my opinion of the chip suggests that, unfortunately, it doesn't take a lot of effort to realize how bad it really is. In a particular light, it may look a lot better, or seem to until you cast that exact same light on, for example, an i5-2500K. I just hope AMD can fix it.

 

[statikregimen]

It takes a serious overclock to make it competitive even with the Phenom II in light- and single-threaded apps. That's about the only light there is for it.....As I've said many times around the net, I'm happy with the chip - I don't mind the OC, as power isn't a concern for me, and since I'm able to get the performance I personally need out of it, and didn't have to switch platforms, then that's good for me. But 99% of others out there will not see that as acceptable, which is bad for AMD.

Has everyone heard, by now, about the B3 stepping? Let's hope the at least address IPC....I think everybody could forgive all of its other faults, if the chip's per-core performance was 25-50% higher. And that is a MASSIVE demand for only a stepping revision.... http://news.softpedia.com/news/AMD-Bulldozer-B3-Revision-Is-in-the-Works-228347.shtml

 

[Onus]

I've decided that whatever I may hope is simply not relevant; I'll believe it when I see it. I will probably rebuild my wife's PC mid-winter to early Spring. If Bulldozer has been fixed, that's what she'll get. If not, most likely it will be an i5-2400.

 

[statikregimen]

Bah, don't be silly! The whole world is built on hopes and dreams :lol:

 

[Onus]

Oh of course! I'm just saying that what I'm going to actually do isn't going to be based on my hopes, but what the reality is in another four months.

 

[statikregimen]

Well, I've been continuously following the situation with BD...Not much new, but I found this guy's videos. Appears he's using mostly ASUS motherboards and his scores - at way higher clocks - are coming in lower than mine...I'll spare too much detail and boring reading, and you can just watch for yourself:

His channel: http://www.youtube.com/user/Realtech4u

and one of the most startling videos, in my opinion:

http://www.youtube.com/watch?v=6qy-z6vlV2U

Now here's my score on an 8120 @ 4.5ghz w/ the MSI 990fxa-gd80 - notice both of his chips are clocked 300mhz faster than mine at 4.8ghz vs. 4.5ghz, and the only one beating me is the 8150 at a meager .04 points:

http://apcommunity.org/images/statik-rig/Cinebench-8120-4500mhz-8c.png

Of course RAM and other factors may contribute, but his frequency/performance scaling seems waaay off by a solid 10% or more. At 4.7ghz, I score 7.71 in cinebench...

All of that to say, maybe there is some weight to motherboard/bios affecting performance in a big way, afterall....We can only hope, wait and see.

 

[jimmysmitty]

this is what he said "Almost all reviews you have seen were fake and based off unavailable CPUS that were recalled. I bought mine from NEWEGG!!!", answering to this : Here are some facts about 95% of the benchmarks for Bulldozer:

1. The CPU used is the FX-8150.

2. They used ASUS Motherboards.

3. The processor they used was an engineering sample.

I was just kidding. But the guy is crazy. These were kits sent out by AMD themselves. They sent out the FX-8150 because thats the current flagship, just like Intel will send out the flagship for SB-E soon. I doubt it was recalled as we would hear something about it.

AMD chose the Asus mobos as they tested it and prefered it.

And all CPUs sent out to reviewers are always ES, which is why when people were saying the results leaked closer to release being on ES were invalid mystified me. They got ES versions of the final B2 FX CPUs. They will have ES versions of the final SB-E CPUs as well. Its normal.

So honestly the guy is either 1. crazy or 2. a AMD fanatic who thinks every site out there is a Intel paid shill site/anti-AMD.

I found one video link where a person said the FX-8150 was the world first native 8 core. I guess they forgot about Beckton....

 

[statikregimen]

I was just kidding. But the guy is crazy. These were kits sent out by AMD themselves. They sent out the FX-8150 because thats the current flagship, just like Intel will send out the flagship for SB-E soon. I doubt it was recalled as we would hear something about it.

AMD chose the Asus mobos as they tested it and prefered it.

And all CPUs sent out to reviewers are always ES, which is why when people were saying the results leaked closer to release being on ES were invalid mystified me. They got ES versions of the final B2 FX CPUs. They will have ES versions of the final SB-E CPUs as well. Its normal.

So honestly the guy is either 1. crazy or 2. a AMD fanatic who thinks every site out there is a Intel paid shill site/anti-AMD.

I found one video link where a person said the FX-8150 was the world first native 8 core. I guess they forgot about Beckton....

Well, I don't want to sound like I'm disagreeing, BUT there is also the chance that Asus and AMD had struck up a deal...Money talks. Obviously, if you look at my post just a few hours ago, you'll see that clearly there can be a serious discrepancy from test bench to test bench. Some of the reviews of the FX chips have been fairly positive (in fact, my personal experience has been largely positive), while others are just depressing to read, and if you notice, usually the motherboards (specifically) are different, in addition to other components, between the ones that read more positively and show better numbers vs those that paint it as a complete failure.

I think we can definitely settle on the fact that performance is overall disappointing (really, no question there), but I think there is something going on - either bug in certain CPUs, BIOS, motherboards or maybe software configuration/compatibility - that is causing some specific test setups to come out way below potential in benchmarks, while others are at least not so abysmal.

As far as the motherboards.org guy is concerned, idk what possessed him to make those comments, but clearly he saw the truth when he benchmarked it for himself (watch his video http://www.youtube.com/watch?v=YOho_YpVF9I).

 

[gnomio]

Its Amds own fault calling it 4 modules/8 Cores and not 4 modules/8 threads which it shouldve been. Now people are benching it core vs core rather than thread to thread.

 

[metallifux]

weren't we supposed to get a motherboard comparison article? In any case my build is finally gonna happen over Xmas and unless there is some revelation i'm converting to intel.

 

[gnomio]

Buy whatever fits your pocket. No one gets any discount from being a dedicated fan we all pay the same price