• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Official Bulldozer PPD??

16K PPD on 6061 bench, 20K PPD on 2684 bench. Breeze was kind enough to play with it last night :D. That was on windows though since the bench is windows, and I do not know other details.
 
Untitledone said:
16K PPD on 6061 bench, 20K PPD on 2684 bench. Breeze was kind enough to play with it last night :D. That was on windows though since the bench is windows, and I do not know other details.
Click to expand...

that was 4.6GHz, 233 HTRef Clock, and 2000ish MHz ram.

At 4.4GHz, 200 HT ref, 1600MHz ram GPU tracker is saying 12-13K, which is what my 4GHz Thuban on windows does. I will probably try bigadv when i get a high enough overclock with reasonable thermals and power consumption. I was able to get it to 5ish GHz, but the power consumption with my 580 while folding was almost 750.
 
Geez so basically you're seeing no improvement over thuban. Actually its worse. That's what I was afraid if.
 
Zero82z said:
So it's slightly worse than an i7 920 running 600MHz slower.
Click to expand...

Do we have recent numbers on quad I7s? As i recall, 20K ppd was before the points adjustment on bigadv. They should be quite a bit worse now.

I'll get some numbers on various OS configurations for BD. My 8120 should be here tomorrow.
 
i7-920 @4.0 will do 18k-20k on regular smp 22k to 26k on bigadv in windows
 
im only running 3.8 now as i have a weak power supply soo long as im not watching movies and surfing im still getting 26k ppd.
 
lehmann said:
im only running 3.8 now as i have a weak power supply soo long as im not watching movies and surfing im still getting 26k ppd.
Click to expand...
Sounds legit, that's still a ton of performance. They do put out a lot of heat at 4ghz but I hear Bulldozer is toasty overclocked as well.
 
Zink said:
Sounds legit, that's still a ton of performance. They do put out a lot of heat at 4ghz but I hear Bulldozer is toasty overclocked as well.
Click to expand...
This. Those who bought 920's around launch have gotten probably the best value for money of any modern cpu, better even than the legendary Q6600's. The fact that 3 years in they are still competitive with AMD's high end is a testament to their design.

The are hot little bastards though.
 
Are they really getting much more than 20k ppd?
 
On my 920, I get about 22-25k PPD on bigadv units depending on how much I use my PC, and around 21-22k on 2684s if I remember correctly. It's been a long time since I ran regular SMP, so I don't remember any numbers for that.
 
I guess I sold my 920 back when 2684 was the only bigadv unit around. I remember barely getting 20k ppd on it, so I would have expected less with the points nerf a couple months back.
 
jebo_4jc said:
I guess I sold my 920 back when 2684 was the only bigadv unit around. I remember barely getting 20k ppd on it, so I would have expected less with the points nerf a couple months back.
Click to expand...
Remember that its points value was also increased.
 
Well I have an I7 -970 and fairly consistantly at 47K ppd on Bigadv units. Thats running right at 4ghz on Windows 7 64bit. It is also running one GTX-460 that puts out another 9.5K ppd.

Others have said to turn off the GPU but I have found that the ppd only goes up to approx 50K. Losing the 9K the GPU turns out just doesn't make sense to me.
 
For comparison purposes, some numbers from my i7.

930 @ 3.8 GHz, 6 GB Ram @ 1600, 11-11-11-29


TPF data from musky's benchmark:


Frame 1: 47 minutes, 17 seconds
Frame 2: 47 minutes, 17 seconds
Frame 3: 47 minutes, 12 seconds
Frame 4: 47 minutes, 8 seconds
Frame 5: 47 minutes, 6 seconds
Frame 6: 47 minutes, 12 seconds
Frame 7: 47 minutes, 24 seconds
Frame 8: 47 minutes, 26 seconds
Frame 9: 47 minutes, 11 seconds
Unit: 2684
Frames: 9
Total run time: 425 minutes, 13 seconds
Average frame time: 47 minutes, 14 seconds


Frame 1: 35 minutes, 58 seconds
Frame 2: 35 minutes, 57 seconds
Frame 3: 36 minutes, 1 seconds
Frame 4: 36 minutes, 15 seconds
Frame 5: 36 minutes, 4 seconds
Frame 6: 36 minutes, 0 seconds
Frame 7: 35 minutes, 58 seconds
Frame 8: 36 minutes, 3 seconds
Frame 9: 35 minutes, 58 seconds
Unit: 2685
Frames: 9
Total run time: 324 minutes, 14 seconds
Average frame time: 36 minutes, 1 seconds


Frame 1: 34 minutes, 9 seconds
Frame 2: 34 minutes, 5 seconds
Frame 3: 34 minutes, 8 seconds
Frame 4: 34 minutes, 9 seconds
Frame 5: 34 minutes, 8 seconds
Frame 6: 34 minutes, 9 seconds
Frame 7: 34 minutes, 7 seconds
Frame 8: 34 minutes, 14 seconds
Frame 9: 34 minutes, 10 seconds
Unit: 2686
Frames: 9
Total run time: 307 minutes, 19 seconds
Average frame time: 34 minutes, 8 seconds


Frame 1: 34 minutes, 0 seconds
Frame 2: 33 minutes, 58 seconds
Frame 3: 33 minutes, 55 seconds
Frame 4: 33 minutes, 58 seconds
Frame 5: 33 minutes, 56 seconds
Frame 6: 34 minutes, 0 seconds
Frame 7: 33 minutes, 58 seconds
Frame 8: 34 minutes, 2 seconds
Frame 9: 34 minutes, 3 seconds
Unit: 6900
Frames: 9
Total run time: 305 minutes, 50 seconds
Average frame time: 33 minutes, 58 seconds


Frame 1: 3 minutes, 10 seconds
Frame 2: 3 minutes, 10 seconds
Frame 3: 3 minutes, 9 seconds
Frame 4: 3 minutes, 10 seconds
Frame 5: 3 minutes, 10 seconds
Frame 6: 3 minutes, 9 seconds
Frame 7: 3 minutes, 15 seconds
Frame 8: 3 minutes, 8 seconds
Frame 9: 3 minutes, 10 seconds
Unit: 6061
Frames: 9
Total run time: 28 minutes, 31 seconds
Average frame time: 3 minutes, 10 seconds


Frame 1: 7 minutes, 37 seconds
Frame 2: 7 minutes, 35 seconds
Frame 3: 7 minutes, 37 seconds
Frame 4: 7 minutes, 37 seconds
Frame 5: 7 minutes, 37 seconds
Frame 6: 7 minutes, 37 seconds
Frame 7: 7 minutes, 37 seconds
Frame 8: 7 minutes, 38 seconds
Frame 9: 7 minutes, 37 seconds
Unit: 6701
Frames: 9
Total run time: 68 minutes, 32 seconds
Average frame time: 7 minutes, 36 seconds




Resulting production:


2684 - 21,683.84 PPD

2685 - 22,794.06 PPD

2686 - 24,706.36 PPD

6900 - 24,888.43 PPD

6061 - 16,556.58 PPD

6701 - 11,913.81 PPD


Maximum power draw running bench was under 300W.
 
While we all wish AMD would have die shrunk Thuban, they DID die shrink their k10.5 quad core in the form of llano. Until now this has been moot because they were bad clockers but they just released the Athlon x4 631 with the gpu disabled.
http://hardforum.com/showthread.php?t=1643127

Early indicators show this chip hitting 4.4-4.7ghz on STOCK VOLTS. Please AMD, glue two of these together Magny Cours style and we will forget about BD for a while...;)
 
R-Type said:
While we all wish AMD would have die shrunk Thuban, they DID die shrink their k10.5 quad core in the form of llano. Until now this has been moot because they were bad clockers but they just released the Athlon x4 631 with the gpu disabled.
http://hardforum.com/showthread.php?t=1643127

Early indicators show this chip hitting 4.4-4.7ghz on STOCK VOLTS. Please AMD, glue two of these together Magny Cours style and we will forget about BD for a while...;)
Click to expand...

God that CPU makes me so [H]ard! AMD needs to do stuff like that more often... I have been drooling over it for a few days now.
 
Linden said:
Are you sure? Isn't that exactly what they did with Bulldozer?
Click to expand...

yup, but the idea still has plenty of room to grow, give it a couple revisions and we will probably see what all the disappointment and suffering was for.


R-Type said:
While we all wish AMD would have die shrunk Thuban, they DID die shrink their k10.5 quad core in the form of llano. Until now this has been moot because they were bad clockers but they just released the Athlon x4 631 with the gpu disabled.
http://hardforum.com/showthread.php?t=1643127

Early indicators show this chip hitting 4.4-4.7ghz on STOCK VOLTS. Please AMD, glue two of these together Magny Cours style and we will forget about BD for a while...;)
Click to expand...

if they were to do it, they would have to to disable one of the IMC's in the process and its probably not an easy thing to do. that and its not really the most power efficient thing in the world. because you are now having to power 2 chips on one socket instead of a single chip, it becomes a wash between it and bulldozer.
 
Last edited:
sirmonkey1985 said:
if they were to do it, they would have to to disable one of the IMC's in the process and its probably not an easy thing to do. that and its not really the most power efficient thing in the world. because you are now having to power 2 chips on one socket instead of a single chip, it becomes a wash between it and bulldozer.
Click to expand...

What I described is exactly how they did Magny Cours and that works very well, in fact 12 core Opteron's only have 2.3B transistors versus the 8 core BD with 2B so the dual Stars would certainly be smaller.

However, unless they already had it in work, there is no way they could start now and beat Piledriver so its largely an academic argument.
 
sirmonkey1985 said:
its not really the most power efficient thing in the world. because you are now having to power 2 chips on one socket instead of a single chip
Click to expand...
Kentsfield vs. Agena power consumption is proof that you're wrong about that
 
Looks like Bulldozer B3 stepping is in work. http://www.anandtech.com/show/4997/amd-working-on-bulldozer-b3-stepping

The rumor sites were reporting that they went through several steppings rapid fire prior to launch that added over 1ghz to the clock rate before they launched with B2. Hopefully this new stepping could come soon and correct some of the issues and get the chip back to design targets (equivalent Thuban IPC, 30% higher clockspeed.
 
R-Type said:
Looks like Bulldozer B3 stepping is in work. http://www.anandtech.com/show/4997/amd-working-on-bulldozer-b3-stepping

The rumor sites were reporting that they went through several steppings rapid fire prior to launch that added over 1ghz to the clock rate before they launched with B2. Hopefully this new stepping could come soon and correct some of the issues and get the chip back to design targets (equivalent Thuban IPC, 30% higher clockspeed.
Click to expand...
It's highly unlikely that there will be any major differences.
 
Zero82z said:
It's highly unlikely that there will be any major differences.
Click to expand...

Of course, just like the complete lack of noticeable difference between B2 and G0 Q6600's, C2 and D0 i7 920's, B2 to B3 Phenoms, or as many of our Extra Spicy folders can attest, the myriad of pre-production Xeon steppings. :rolleyes:

Steppings can and do make a big difference and provide an opportunity for the designers to refine the design and fix problems that were discovered after the launch stepping was taped out. Particuarly given AMD knew B2 BD underperformed, I'll bet they left much more of the team on to complete this 'sustaining' work than normal.

I'm not sure where your view on this originates from, but I don't see any basis for it given what we've seen over the last five years. :confused:
 
No stepping changes as long as I can remember have ever resulted in significant performance improvements. If you can present an example to the contrary, I'd love to see it. Better overclocking is a possibility, but that's about it. And as far as pre-production steppings are concerned, that is a totally different situation.
 
Zero82z said:
No stepping changes as long as I can remember have ever resulted in significant performance improvements. If you can present an example to the contrary, I'd love to see it. Better overclocking is a possibility, but that's about it. And as far as pre-production steppings are concerned, that is a totally different situation.
Click to expand...

B2 to G0 Q6600 stepping revision increased temperature limits by 10C, lowered power draw substantially, ran much cooler, and generally overclocked around 400mhz higher.

C2 to D0 Nehalem is a similar story

B2 to B3 Phenom 1, resolved TLB bug, 64 bit issues, power draw, and added over 500mhz to the clock ceiling. These bugs aren't that much different in complexity to some of what haunts BD.


And in any case, were they not rushed I think B2 would have stayed a pre-production stepping as they fixed these bugs before launch. The change to B3 could be as large as any pre-production move.
 
R-Type said:
Looks like Bulldozer B3 stepping is in work. http://www.anandtech.com/show/4997/amd-working-on-bulldozer-b3-stepping

The rumor sites were reporting that they went through several steppings rapid fire prior to launch that added over 1ghz to the clock rate before they launched with B2. Hopefully this new stepping could come soon and correct some of the issues and get the chip back to design targets (equivalent Thuban IPC, 30% higher clockspeed.
Click to expand...

like zero said, very unlikely. at best with a new stepping we will see better power usage and higher overclocks. if they could get the power usage to a controllable level i think that would at least make it a more viable choice even though the performance isn't there.

bulldozer B3 stepping will be the same as phenom II C3 stepping is to phenom II C2 stepping basically.


R-Type said:
And in any case, were they not rushed I think B2 would have stayed a pre-production stepping as they fixed these bugs before launch. The change to B3 could be as large as any pre-production move.
Click to expand...

B2 is the updated stepping to bulldozer, B1 was the preproduction stepping which failed to meet AMD's clock requirements so it was scrapped and B2 stepping went into production which pushed the release date back.
 
Last edited:
You are discounting increased overclocking headroom like it is a non factor, who here is honestly going to be running these stock? Overclocking headroom IS performance for enthusiast users.
 
R-Type said:
B2 to G0 Q6600 stepping revision increased temperature limits by 10C, lowered power draw substantially, ran much cooler, and generally overclocked around 400mhz higher.

C2 to D0 Nehalem is a similar story

B2 to B3 Phenom 1, resolved TLB bug, 64 bit issues, power draw, and added over 500mhz to the clock ceiling. These bugs aren't that much different in complexity to some of what haunts BD.


And in any case, were they not rushed I think B2 would have stayed a pre-production stepping as they fixed these bugs before launch. The change to B3 could be as large as any pre-production move.
Click to expand...
And like I said, in none of those cases was there any significant performance increase. Also, you're wrong about Nehalem. There was no C2 stepping, only C0/C1; the TDP did not change at all, and overclocking headroom was slightly improved, but only by about 100-200MHz on average. The change was not particularly dramatic.
R-Type said:
You are discounting increased overclocking headroom like it is a non factor, who here is honestly going to be running these stock? Overclocking headroom IS performance for enthusiast users.
Click to expand...
So maybe they'll be able to reach 4.8GHz instead of 4.6GHz. Even so, they'll still be slower than Sandy Bridge, while consuming considerably more power. There won't be any miracles worked.
 
Last edited:
You must log in or register to reply here.
Back
Top