Fudzilla: Bulldozer performance figures are in

Page 45 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Status
Not open for further replies.

lau808

Senior member
Jun 25, 2011
217
0
71
are we there yet? ugh, the suspense is killing me!
heres my prediction

2600k fx8100
fx6100
2500k
fx4100

anyone know how much better ipc is 2500k vs phenom II x4?
 

lau808

Senior member
Jun 25, 2011
217
0
71
ok roughly 40% as he said, wow i knew it was bad but yuck. hope bd can bring ipc closer by half would be nice. 20% would be much nicer though still losing not to mention ivy coming out after.
 

Crap Daddy

Senior member
May 6, 2011
610
0
0
I think the 8150 will struggle with 8 cores, higher TDP and clocks to compete with the 2600K but it will be close. As for the 2500K, I don't think that AMD can put something on the market at that price point to compete.
 

frostedflakes

Diamond Member
Mar 1, 2005
7,925
1
81
AMD's TDP-based turbo sounds pretty cool, seems to be more sophisticated than Intel's turbo technology. Other than that, though, it seems pretty lackluster compared to Intel's offerings. The article hints that IPC may be *lower* than current K10.5 chips due to the shared resources approach BD is taking.
 

nonameo

Diamond Member
Mar 13, 2006
5,902
2
76
AMD's TDP-based turbo sounds pretty cool, seems to be more sophisticated than Intel's turbo technology. Other than that, though, it seems pretty lackluster compared to Intel's offerings. The article hints that IPC may be *lower* than current K10.5 chips due to the shared resources approach BD is taking.

FYI, it's the same type of turbo being used in llano
 

Mopetar

Diamond Member
Jan 31, 2011
8,496
7,752
136
I don't know why this lower IPC idea keeps cropping up. We've had an AMD rep post time and time again, that IPC is going to be increasing. There's been no indication of how much it will increase, but we know that it's not decreasing.
 

frostedflakes

Diamond Member
Mar 1, 2005
7,925
1
81
Have we? Reps have said that performance would be increasing (by 50% compared to current products? or 35%, something like that). Not surprising considering core count and clock speeds are increasing with BD, as these can compensate for lower IPC. AFAIK nobody has made claims about IPC increasing, though, but I may be mistaken.
 

cantholdanymore

Senior member
Mar 20, 2011
447
0
76
JFAMD said in a post that IPC compared to current AMD processors is increased. IPC is reduced compared to an "hypothetical" bulldozer core that doesn't share resources. Don't ask me to prove it, the post is some where in this forum.
 

lau808

Senior member
Jun 25, 2011
217
0
71
yea, 1 module with both cores active will equal 180% performance of a module with 1 core active. which is y i assume 4 core will be below 2500k unless they can have 10% greater ipc (lol) than sb. but i also understand that its way better than the 2600k's ht which is 120-140% of sb core. so fx8100 needs to be 80% of sb ipc to match clock for clock give or take which is possible imo. and if not, then the slightly higher stock clock should help it be about even in 8 thread performance vs 2600k
 

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
Thuban - 346 mm2
Bulldozer - 315mm2
Sandy Bridge - 216mm2

It is definitely smaller than the Phenom II X6 but SB is 99mm2 smaller and a portion of the SB die is dedicated to an IGP. If it barely matches the Core i7 2600K in terms of performance then it's not good news for BD. D:
 

PreferLinux

Senior member
Dec 29, 2010
420
0
0
yea, 1 module with both cores active will equal 180% performance of a module with 1 core active. which is y i assume 4 core will be below 2500k unless they can have 10% greater ipc (lol) than sb. but i also understand that its way better than the 2600k's ht which is 120-140% of sb core. so fx8100 needs to be 80% of sb ipc to match clock for clock give or take which is possible imo. and if not, then the slightly higher stock clock should help it be about even in 8 thread performance vs 2600k
No, it will be 111% by my calculations.
 

lau808

Senior member
Jun 25, 2011
217
0
71
No, it will be 111% by my calculations.
how is that? a few pages back we spoke about this (myself and someone else) amd said 1 module will have 180% performance of 1 core. and someone else later said ht is between 120-140% i'd link it but i dont remember where. i'll look back a few pages in a bit tho
 

Riek

Senior member
Dec 16, 2008
409
15
76
Thuban - 346 mm2
Bulldozer - 315mm2
Sandy Bridge - 216mm2

It is definitely smaller than the Phenom II X6 but SB is 99mm2 smaller and a portion of the SB die is dedicated to an IGP. If it barely matches the Core i7 2600K in terms of performance then it's not good news for BD. D:

If it is only able to tie 2500 for example.... it is still alot better then what they have now! It is smaller and alot more competitive than their current lineup. If they would have a die size problem they would make masks for the FX6 and FX4. But they don't because their volume for BD isn't big enough to worry about that.
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
BD single module, single thread = 100%
BD single module, dual thread = 180%

BD 4 module should be 180% * 4 = 720
Thuban 6C = 6*100 = 600

(Those percentages above are theoretical Maximum)

I expect an average 20% higher IPC than Thuban

Take 6Core Thuban (1100T at 3.3GHz) and we add 9% more frequency to get to 3.6GHz the BD FX8150 will be.

BD 4 module should be (180% + 29%) * 4 = 928

If the above maths are correct (I was never good with maths) then BD FX8150 will be close to ~55% faster than Thuban 1100T in multithreaded apps (no turbo both of them)
 

lau808

Senior member
Jun 25, 2011
217
0
71
imo, if bd8core matches 2600k its def a win. amd will be competitive again. granted sb will still have better ipc, they will have a cpu to compete with 2600k and 2500k.

an even bettter situation would be bd4 matching/beating 2500k and bd6 matching/beating 2600k and bd8 to 6 core sb-e but im pretty sure that aint happening.
 
Status
Not open for further replies.