Overclock.net › Forums › AMD › AMD CPUs › Bulldozer Live testing in China
New Posts  All Forums:Forum Nav:

Bulldozer Live testing in China - Page 26

post #251 of 275
Quote:
Originally Posted by Phantom123 View Post
Lets not forget, according to Semi-Accurate the server processors release date has been switched with desktops. So maybe we will get to see the server bulldozers in July and can just make a math problem out of the gigahertz to see how the desktops will perform.
Excellent thought, it always better to evaluate a production chip and it becomes a math problem most times if the scaling is linear.
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
post #252 of 275
Quote:
Originally Posted by BallaTheFeared View Post
The difference is, like with current gen products and the new AMD chip coming out - is that the 950 can disable HT and retain ~85% of its performance, whereas disabling 2- or 4 cores in this case now results in a much larger performance loss. Up to we can now presume will be 50% with the new bulldozer x8 chip.

I'm gonna bow out here, I'll wait for the next round of leaks to assess the situation again. My only interest is in the x4 chip anyways since its the only one that will be priced near my i5-2500k. And as a gamer the only core amount that matters to me currently is four (4). We'll see if Bulldozer has the same per core performance as Intel then and I can always sell and switch over at that time. If that is true then the x8 bulldozer chip should score around what... 11 at 3.4GHz, or closer to actually 14 maybe 15 at stock.

We'll see what happens though!
Another straw man argument. the architectures are different, you will not disable the other core in a module.

Yes if the math is correct you will get around 11 at 3.4. Under the assumption that it scales linear. From the other discussions theads the second int core is only worth about 1.8 vs 1.2 or 1.3 for HT. This needs to be verified with a production chips.
Edited by Kmon - 6/11/11 at 7:27pm
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
post #253 of 275
Quote:
Originally Posted by BallaTheFeared View Post
The difference is, like with current gen products and the new AMD chip coming out - is that the 950 can disable HT and retain ~85% of its performance, whereas disabling 2- or 4 cores in this case now results in a much larger performance loss. Up to we can now presume will be 50% with the new bulldozer x8 chip.

I'm gonna bow out here, I'll wait for the next round of leaks to assess the situation again. My only interest is in the x4 chip anyways since its the only one that will be priced near my i5-2500k. And as a gamer the only core amount that matters to me currently is four (4). We'll see if Bulldozer has the same per core performance as Intel then and I can always sell and switch over at that time. If that is true then the x8 bulldozer chip should score around what... 11 at 3.4GHz, or closer to actually 14 maybe 15 at stock.

We'll see what happens though!
You just proved the point that HT is inefficient as it is only about 1.15 in your example. It is really about 1.2 to 1.3 effective of a full intel core. The BD arhitecture is "estimated" to be 1.80 effective of a full AMD Core. It appears the BD architecture is approx 38 percent more effective than the Intel arhitecture. NOTE I DID NOT SAY CORE PERFORMANCE

4 threads with HT 5.2 times the core performance
4 modules 7.2 times the core performance
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
post #254 of 275
Easy there quad poster. I think 1.8 represents two cores, not one. Which is why I'm hesitant of bulldozer. 1.8 + 1.8 = 3.6 Whereas with the i5 I get 3.98, so unless bulldozer x8 is rocking out at 11+ points at 3.4Ghz, why on earth would I ever spend $200 for the 4 core version?

I don't care for HT, but I don't need eight cores either.

What do I do now? Get the design that wants to use eight cores to compete with 4 /w HT, but only the chopped down 50% less effective model instead so I'm not paying as much for my cpu as I did for my gpus?


Do you see what I'm saying? I'm not sure if I'm being clear enough.

The x8 bulldozer chip will need to reach 11 or better around 3.4Ghz (probably closer to 15 @ stock right? I mean we're talking about something that would kill a highly overclocked 990x on water/phase), and clock to around 4.8GHz 24/7 on average (the 4 core) for me even consider the 4 core version which will be priced right around $200 and in direct completion with the i5-2500k.
Edited by BallaTheFeared - 6/11/11 at 7:45pm
    
CPUMotherboardGraphicsGraphics
Intel Core i5 2500K P8P67 PRO NVIDIA GeForce GTX 470 NVIDIA GeForce GTX 470 
GraphicsRAMRAMRAM
NVIDIA GeForce 9800 GT G-Skill A-Data G-Skill 
RAMHard DriveOptical DriveOS
A-Data Crucial M4 64GB + 1TB F3 Spinpoint $155 LS/DL DVD RW $?? Windows 8 64-bit "Epic Registry" Edition 
MonitorPowerCase
ASUS 21.5 1920x1080 2ms $135 CORSAIR HX850 $120 Mother Earth $free 
  hide details  
Reply
    
CPUMotherboardGraphicsGraphics
Intel Core i5 2500K P8P67 PRO NVIDIA GeForce GTX 470 NVIDIA GeForce GTX 470 
GraphicsRAMRAMRAM
NVIDIA GeForce 9800 GT G-Skill A-Data G-Skill 
RAMHard DriveOptical DriveOS
A-Data Crucial M4 64GB + 1TB F3 Spinpoint $155 LS/DL DVD RW $?? Windows 8 64-bit "Epic Registry" Edition 
MonitorPowerCase
ASUS 21.5 1920x1080 2ms $135 CORSAIR HX850 $120 Mother Earth $free 
  hide details  
Reply
post #255 of 275
Quote:
Originally Posted by Kmon View Post
Do you know what writeback is and what it does?
Quote:
Originally Posted by Kmon View Post
Excellent thought, it always better to evaluate a production chip and it becomes a math problem most times if the scaling is linear.
Quote:
Originally Posted by Kmon View Post
Another straw man argument. the architectures are different, you will not disable the other core in a module.

Yes if the math is correct you will get around 11 at 3.4. Under the assumption that it scales linear. From the other discussions theads the second int core is only worth about 1.8 vs 1.2 or 1.3 for HT. This needs to be verified with a production chips.
Quote:
Originally Posted by Kmon View Post
You just proved the point that HT is inefficient as it is only about 1.15 in your example. It is really about 1.2 to 1.3 effective of a full intel core. The BD arhitecture is "estimated" to be 1.80 effective of a full AMD Core. It appears the BD architecture is approx 38 percent more effective than the Intel arhitecture. NOTE I DID NOT SAY CORE PERFORMANCE

4 threads with HT 5.2 times the core performance
4 modules 7.2 times the core performance

Do you know what an edit button is and what it does?
Horizon
(15 items)
 
ToSHITa
(10 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core I5 2500K Gigabyte GA-B75M-D3H XFX Radeon R9-280 Double Dissipation Corsair Vengance 16GB Dual Channel DDR3 
Hard DriveHard DriveHard DriveCooling
Crucial M500 120GB SSD WD BLACK SERIES WD1003FZEX 1TB 7200 RPM 64MB Ca... Western Digital WD Blue WD10EZEX 1TB 7200 RPM 6... Corsair H100 
OSMonitorMonitorKeyboard
Windows 7 Ultimate Edition Alienware Optx AW2210 Asus VS239 CM Storm QuickFire Rapid Cherry MX Brown 
PowerCaseMouse
NZXT Hale75 750W PSU Fractal Design Node 804 Logitech G700s 
CPUMotherboardGraphicsRAM
AMD Quad-Core A8-4500M Satellite L855D-S5114 Radeon™ HD 7640G Generic 4GB DDR3-1600MHz (PC3-12800) 
Hard DriveOptical DriveOSMonitor
Crucial M4 128GB SSD Generic DVD SuperMulti DVD-RAM (5x) Windows 10 Home (Hardcore Old-key Fresh Install... 15.6" Widescreen TruBrite TFT @ 1366x768 
PowerAudio
65W (19v 3.42a) 100-240/50-60Hz AC Adaptor SRS Premium Sound HD 
CPUMotherboardGraphicsRAM
Pentium Dual T2310 Gateway C Series Intel Accelerated Graphics 1024MB 
Hard DriveOSMonitorPower
80GB Linux Mint 12 Wacom Penabled 65W Power Brick 
  hide details  
Reply
Horizon
(15 items)
 
ToSHITa
(10 items)
 
 
CPUMotherboardGraphicsRAM
Intel Core I5 2500K Gigabyte GA-B75M-D3H XFX Radeon R9-280 Double Dissipation Corsair Vengance 16GB Dual Channel DDR3 
Hard DriveHard DriveHard DriveCooling
Crucial M500 120GB SSD WD BLACK SERIES WD1003FZEX 1TB 7200 RPM 64MB Ca... Western Digital WD Blue WD10EZEX 1TB 7200 RPM 6... Corsair H100 
OSMonitorMonitorKeyboard
Windows 7 Ultimate Edition Alienware Optx AW2210 Asus VS239 CM Storm QuickFire Rapid Cherry MX Brown 
PowerCaseMouse
NZXT Hale75 750W PSU Fractal Design Node 804 Logitech G700s 
CPUMotherboardGraphicsRAM
AMD Quad-Core A8-4500M Satellite L855D-S5114 Radeon™ HD 7640G Generic 4GB DDR3-1600MHz (PC3-12800) 
Hard DriveOptical DriveOSMonitor
Crucial M4 128GB SSD Generic DVD SuperMulti DVD-RAM (5x) Windows 10 Home (Hardcore Old-key Fresh Install... 15.6" Widescreen TruBrite TFT @ 1366x768 
PowerAudio
65W (19v 3.42a) 100-240/50-60Hz AC Adaptor SRS Premium Sound HD 
CPUMotherboardGraphicsRAM
Pentium Dual T2310 Gateway C Series Intel Accelerated Graphics 1024MB 
Hard DriveOSMonitorPower
80GB Linux Mint 12 Wacom Penabled 65W Power Brick 
  hide details  
Reply
post #256 of 275
Quote:
Originally Posted by BallaTheFeared View Post
Easy there quad poster. I think 1.8 represents two cores, not one. Which is why I'm hesitant of bulldozer. 1.8 + 1.8 = 3.6 Whereas with the i5 I get 3.98, so unless bulldozer x8 is rocking out at 11+ points at 3.4Ghz, why on earth would I ever spend $200 for the 4 core version?

I don't care for HT, but I don't need eight cores either.

What do I do now? Get the design that wants to use eight cores to compete with 4 /w HT, but only the chopped down 50% less effective model instead so I'm not paying as much for my cpu as I did for my gpus?


Do you see what I'm saying? I'm not sure if I'm being clear enough.

The x8 bulldozer chip will need to reach 11 or better around 3.4Ghz (probably closer to 15 @ stock right? I mean we're talking about something that would kill a highly overclocked 990x on water/phase), and clock to around 4.8GHz 24/7 on average (the 4 core) for me even consider the 4 core version which will be priced right around $200 and in direct completion with the i5-2500k.

Yeah, you get 7.2 for the 4 module 8 core model, which equals the 2600K HT 4 cores eight threads. Let’s compare apples to apples threads to threads, which is what I was doing.

It depends upon the AMD core performance if it kills a 990x. This we do not know.
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
post #257 of 275
We shall see, apples to apples is the 4000 series x4 vs my i5-2500k. There is more than one bulldozer chip, because there is more than one market segment. When we speak of bulldozer, we should probably include all models instead of only the $300+ end.

Hence when I said at the end of the day Intel need only remove HT from their mainstream SB lineup and drop $100 off the price. Which is about a 20% performance loss in multithreading. However with AMDs new design in order to hit the $200 price segment they removed four "real" cores from their processor, which equates into a 50% performance loss.

We shall see if it equals the i7-2600k, it should, and it should exceed it. It should be able to compete with the 990x at least in multithreading. However I think the ambitions of clock for clock core for core performance equaling sandy bridge are a bit lofty. Basically to do that these samples would need to be running at 40% bulldozers actual capacity. I don't think cache is going to make that big of a difference, is it is indeed defective.
    
CPUMotherboardGraphicsGraphics
Intel Core i5 2500K P8P67 PRO NVIDIA GeForce GTX 470 NVIDIA GeForce GTX 470 
GraphicsRAMRAMRAM
NVIDIA GeForce 9800 GT G-Skill A-Data G-Skill 
RAMHard DriveOptical DriveOS
A-Data Crucial M4 64GB + 1TB F3 Spinpoint $155 LS/DL DVD RW $?? Windows 8 64-bit "Epic Registry" Edition 
MonitorPowerCase
ASUS 21.5 1920x1080 2ms $135 CORSAIR HX850 $120 Mother Earth $free 
  hide details  
Reply
    
CPUMotherboardGraphicsGraphics
Intel Core i5 2500K P8P67 PRO NVIDIA GeForce GTX 470 NVIDIA GeForce GTX 470 
GraphicsRAMRAMRAM
NVIDIA GeForce 9800 GT G-Skill A-Data G-Skill 
RAMHard DriveOptical DriveOS
A-Data Crucial M4 64GB + 1TB F3 Spinpoint $155 LS/DL DVD RW $?? Windows 8 64-bit "Epic Registry" Edition 
MonitorPowerCase
ASUS 21.5 1920x1080 2ms $135 CORSAIR HX850 $120 Mother Earth $free 
  hide details  
Reply
post #258 of 275
Hi Balla;

From the AMD http://www.amdzone.com/phpbb3/viewto...138624#p205857

"Average IPC integer will be higher. If you consider that the average IPC with pure integer code is something about 0.8-1, only logical conclusion is that the K10 ALU/AGU-s are pretty much underutilized. Main idea with Bulldozer is to make it for better utilization of resources.

If you have 2 ALU's with utilization of 60% that's give you IPC of 1.2 on average.
If you have 3 ALU's with utilization of 35% that's give you IPC of 1.05 on average.
Which is better, and more power efficient?

AMD's approach is to make smaller integer core and better utilized. Intel's approach is to make larger integer core and increase utilization with hyperthreading.

However, hyperthreading brings ~25% more utilization of 3-way integer core. If your integer core has 45% utilization with single thread, that is 1.35, per single thread, which is around ~20% faster than K10 for integer ops. With hyperthreaded core, utilization can go up to 1.65 IPC, which is 52-55% of utilization of 3 ALU's in the Sandy B. core.

With Bulldozer and two threads you can go up 60% of utilization of 2 ALU cores, which is 2.4 IPC with two threads per module vs. 1.5-1.6 IPC with two threads per Intel FAT core. That is ~50% faster than FAT core per clock.
Goal is to reach 50% more performance with 33% more ALU's (4 ALU's vs. 3 ALU's) and probably with same power envelope.

With single thread integer IPC will be slightly lower than SB, but not all workloads are integer. There is a lot of mixed code, and there are dedicated load/store and data caches with BD module vs. SB core."

Need to thank Tosh of the link to the discussion that the above was taken from.
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
Gunslinger
(13 items)
 
  
CPUMotherboardGraphicsRAM
1090 MSI 890FXA GD70 5870 Corsair 
Hard DriveOSMonitorPower
C300 Windows I-INC Thermaltake 850 
Case
HAF 932 
  hide details  
Reply
post #259 of 275
I skipped over it to be honest, amdbias.com is the last place I'd take anything at face value.
    
CPUMotherboardGraphicsGraphics
Intel Core i5 2500K P8P67 PRO NVIDIA GeForce GTX 470 NVIDIA GeForce GTX 470 
GraphicsRAMRAMRAM
NVIDIA GeForce 9800 GT G-Skill A-Data G-Skill 
RAMHard DriveOptical DriveOS
A-Data Crucial M4 64GB + 1TB F3 Spinpoint $155 LS/DL DVD RW $?? Windows 8 64-bit "Epic Registry" Edition 
MonitorPowerCase
ASUS 21.5 1920x1080 2ms $135 CORSAIR HX850 $120 Mother Earth $free 
  hide details  
Reply
    
CPUMotherboardGraphicsGraphics
Intel Core i5 2500K P8P67 PRO NVIDIA GeForce GTX 470 NVIDIA GeForce GTX 470 
GraphicsRAMRAMRAM
NVIDIA GeForce 9800 GT G-Skill A-Data G-Skill 
RAMHard DriveOptical DriveOS
A-Data Crucial M4 64GB + 1TB F3 Spinpoint $155 LS/DL DVD RW $?? Windows 8 64-bit "Epic Registry" Edition 
MonitorPowerCase
ASUS 21.5 1920x1080 2ms $135 CORSAIR HX850 $120 Mother Earth $free 
  hide details  
Reply
post #260 of 275
This is hilarious...

The thing isn't released. Quit acting like you know how it performs.

My opinion, if it performs like these benchmarks AMD didn't do so well this time around. If it is better, hooray.
Bench Stuffs
(32 items)
 
ITX Daily Driver
(11 items)
 
For Sale: 6GHz+ 980X
$225 (USD) or best offer
CPUCPUCPUCPU
Intel Core i7 920 Intel Xeon W3520 Intel Core i7 4770K Intel Core 2 Duo E8500 
CPUCPUCPUCPU
Intel Pentium G3258 Intel QX9650 AMD FX 4300 Intel E2140 
CPUMotherboardGraphicsGraphics
E4300 Asus Rampage Extreme / P5E3 Deluxe / Gigabyte X... MSI GTX 275 Lightning MSI HD 7970 Lightning 
GraphicsGraphicsGraphicsGraphics
Asus GTX 285 Matrix Platinum MSI HD 5870 Lightning EVGA GTX 980 TI Classified K|ngp|n Edition MSI GTX 580 Lightning 
RAMRAMRAMRAM
Corsair Dominator Platinum 2666 10-12-12 Samsung Corsair Dominator GT 2000 7-8-7 Hypers Kingston HyperX 2000 9-9-9 BBSE G.Skill Trident 2000 9-9-9 PSC 
Hard DriveHard DriveHard DriveCooling
Kingston SSD Now! 128GB Patriot Torqx 2 32GB Seagate 80GB XPSC Rasa + EK VGA + HWLabs Black Ice GTX 360 +... 
OSOSOSKeyboard
W7 Pro x64 SP1 Windows XP x86 SP3 W8.1 Ducky Shine 3, Cherry Brown Switches 
PowerCaseMouseMouse Pad
Corsair AX1200 Dimastech Easy V2.5! <3 Logitech G5 Battlefield 2142 Edition Puretrak Stealth 
  hide details  
Reply
Bench Stuffs
(32 items)
 
ITX Daily Driver
(11 items)
 
For Sale: 6GHz+ 980X
$225 (USD) or best offer
CPUCPUCPUCPU
Intel Core i7 920 Intel Xeon W3520 Intel Core i7 4770K Intel Core 2 Duo E8500 
CPUCPUCPUCPU
Intel Pentium G3258 Intel QX9650 AMD FX 4300 Intel E2140 
CPUMotherboardGraphicsGraphics
E4300 Asus Rampage Extreme / P5E3 Deluxe / Gigabyte X... MSI GTX 275 Lightning MSI HD 7970 Lightning 
GraphicsGraphicsGraphicsGraphics
Asus GTX 285 Matrix Platinum MSI HD 5870 Lightning EVGA GTX 980 TI Classified K|ngp|n Edition MSI GTX 580 Lightning 
RAMRAMRAMRAM
Corsair Dominator Platinum 2666 10-12-12 Samsung Corsair Dominator GT 2000 7-8-7 Hypers Kingston HyperX 2000 9-9-9 BBSE G.Skill Trident 2000 9-9-9 PSC 
Hard DriveHard DriveHard DriveCooling
Kingston SSD Now! 128GB Patriot Torqx 2 32GB Seagate 80GB XPSC Rasa + EK VGA + HWLabs Black Ice GTX 360 +... 
OSOSOSKeyboard
W7 Pro x64 SP1 Windows XP x86 SP3 W8.1 Ducky Shine 3, Cherry Brown Switches 
PowerCaseMouseMouse Pad
Corsair AX1200 Dimastech Easy V2.5! <3 Logitech G5 Battlefield 2142 Edition Puretrak Stealth 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD CPUs
Overclock.net › Forums › AMD › AMD CPUs › Bulldozer Live testing in China