Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [YT] AMD FX Bulldozer @ 6.5 GHz - under LN2
New Posts  All Forums:Forum Nav:

[YT] AMD FX Bulldozer @ 6.5 GHz - under LN2 - Page 8

post #71 of 116
Quote:
Originally Posted by gplnpsb;14680580 
OBR has tried very hard to hide the model number of his precious engineering sample, but it looks like he's now slipped up. At about 7 seconds into the video you can just make out the model number of the CPU: "AMD Eng Sample, ZD2851W8K44". The digits K44 at the end indicate that this is the old OR-B0 stepping. The other digits indicate that it is a desktop engineering sample with eight cores and a 2.8 GHz stock clock. It will be interesting to see how the retail chips (presumably B2 or B3 stepping) will clock compared to this B0 sample.

zd28251w8k44.jpg

DUDE!!!
EXCELLENT CSI WORK!

+1 for attention to detail.
BLACK!CE
(13 items)
 
  
CPUMotherboardGraphicsRAM
2500k Z68 Exetreme4 Gen3 MSI 6950 Twin Frozr II 8g 8-9-8 24 1600 
Power
CM 800w Silent Pro Gold 
  hide details  
Reply
BLACK!CE
(13 items)
 
  
CPUMotherboardGraphicsRAM
2500k Z68 Exetreme4 Gen3 MSI 6950 Twin Frozr II 8g 8-9-8 24 1600 
Power
CM 800w Silent Pro Gold 
  hide details  
Reply
post #72 of 116
Quote:
Originally Posted by Nutty;14681545 
That was some pretty crazy inaccurate math. Even if it was somehow "accurate", SuperPi is a horrible way of measuring performance as it favors performance towards Intels CPUs. That makes it impossible to compare the performance of AMD and Intel against each other in this benchmark.

It favors Intel because it's a single thread benchmark app and AMD cores are slower than Intel's. However, when you throw twice the amount on the die than Intel does...you can dominate them in threaded apps as long as your cores are slightly better than half as good.

Please show me how my math was inaccurate though because I'm honestly trying to derive some sense from this otherwise useless video.

6.5 GHz = 8 seconds.
6.5 / 5 = 1.3
5 GHz = 10.4 seconds. (GHz increases scale nearly linearly in every chip arch I've come across)

10.4 / 7.5 (the 2500k score @ 5 GHz) = 1.387
1/1.387 = 0.72
= 72%

So one BD core is 38.7% slower than a SB core; or worth 72% of a SB core. Same thing.

8 cores x 0.72 = 5.8

Thus, 8 BD cores is equivalent to ~5.8 Sandy Bridge cores @ same clock speed. Or you can just say BD is 45% faster than SB.

I don't know how SuperPi interacts with Intel's Hyperthreading (because it's a totally useless bench for any real world purposes) but we can try to draw a couple conclusions here.

1) If BD = 45% faster than 2500k, we might see CineBench scores of 11.8 @ 5 GHz.

2) If BD = 45% faster than 2600k, we might see CineBenches of > 14 @ 5 GHz.

If either of these is true at all, I'll be on 1). A CPU that could pull 14pts in CineBench is worth at least $1,300...so going based on the estimated price of $320, I'd say even a score of 11.8 would be too good to be true, but it could be possible since BD probably won't be amazing at apps that don't use all 8 cores.
Edited by kweechy - 8/22/11 at 11:09pm
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
post #73 of 116
Quote:
Originally Posted by Darkapoc;14679437 
What a let down, I was all giddy and then it just stopped before showing me the super pi time frown.giffrown.giffrown.gif

its obr's MO


Quote:
Originally Posted by kweechy;14681491 
MATH TIME!!!!!!!!!

If Bulldozer is 8 seconds @ 6.5 GHz, it's roughly 10.5 seconds @ 5 GHz.
Let's say that an i5 2500k is 7.5 seconds @ 5 GHz.

This means Bulldozer's power per core is roughly 72% of a Sandy Bride core...then AMD gives us EIGHT of those cores to play with over the 4 from Intel.

8 cores @ 72% = 5.8 equivalent SB cores; 45% faster than a 2500k in fully threaded apps. Bulldozer might actually come pretty damn close to matching SB-E let alone SB...of course that's if this video and thread is worth anything at all.

I'm betting it's not.


its an old bd es cpu .. i dont know why are people in this thread still taking the time to do some math on things like that


superpi is just wrong in term of gauging difference ...

there are other single threaded benchmarks that are more accurate
My System
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX 8120 Asus crosshair V formula 2 msi HD6870 Twin Frozr II CORSAIR Vengeance 8GB 
Hard DriveOptical DriveOSMonitor
2 SAMSUNG F4 2TB boot + 4 WD 2TB storage Lite On blu ray player win 7 sp1 oem Asus VH236H @ 1920 x 1080 
KeyboardPowerCaseMouse
logitech Media Elite Keyboard ENERMAX MAXREVO 1350W psu SILVERSTONE RAVEN RV02B-EW Matte black logitech OEM 
  hide details  
Reply
My System
(13 items)
 
  
CPUMotherboardGraphicsRAM
FX 8120 Asus crosshair V formula 2 msi HD6870 Twin Frozr II CORSAIR Vengeance 8GB 
Hard DriveOptical DriveOSMonitor
2 SAMSUNG F4 2TB boot + 4 WD 2TB storage Lite On blu ray player win 7 sp1 oem Asus VH236H @ 1920 x 1080 
KeyboardPowerCaseMouse
logitech Media Elite Keyboard ENERMAX MAXREVO 1350W psu SILVERSTONE RAVEN RV02B-EW Matte black logitech OEM 
  hide details  
Reply
post #74 of 116
Quote:
Originally Posted by kweechy;14681834 
It favors Intel because it's a single thread benchmark app and AMD cores are slower than Intel's. However, when you throw twice the amount on the die than Intel does...you can dominate them in threaded apps as long as your cores are slightly better than half as good.

Please show me how my math was inaccurate though because I'm honestly trying to derive some sense from this otherwise useless video.

6.5 GHz = 8 seconds.
6.5 / 5 = 1.3
5 GHz = 10.4 seconds. (GHz increases scale nearly linearly in every chip arch I've come across)

10.4 / 7.5 (the 2500k score @ 5 GHz) = 1.387
1/1.387 = 0.72
= 72%

So one BD core is 38.7% slower than a SB core; or worth 72% of a SB core. Same thing.

8 cores x 0.72 = 5.8

Thus, 8 BD cores is equivalent to ~5.8 Sandy Bridge cores @ same clock speed. Or you can just say BD is 45% faster than SB.

I don't know how SuperPi interacts with Intel's Hyperthreading (because it's a totally useless bench for any real world purposes) but we can try to draw a couple conclusions here.

1) If BD = 45% faster than 2500k, we might see CineBench scores of 11.8 @ 5 GHz.

2) If BD = 45% faster than 2600k, we might see CineBenches of > 14 @ 5 GHz.

If either of these is true at all, I'll be on 1). A CPU that could pull 14pts in CineBench is worth at least $1,300...so going based on the estimated price of $320, I'd say even a score of 11.8 would be too good to be true, but it could be possible since BD probably won't be amazing at apps that don't use all 8 cores.

That is NOT the reason. It favors Intel because SPi uses an outdated, rather obsolete instruction set called x87. x87 has basically been replaced by SSE (and it variants) in modern applications. However, Intel still uses x87 on their CPUs wheras AMD just emulates it... this is the main reason why SPi is faster on Intel processors.

If you want to compare single threaded performance between AMD and Intel there are several better and more representative benchmarks that you can use.
post #75 of 116
Quote:
Originally Posted by JCPUser;14681899 
That is NOT the reason. It favors Intel because SPi uses an outdated, rather obsolete instruction set called x87. x87 has basically been replaced by SSE (and it variants) in modern applications. However, Intel still uses x87 on their CPUs wheras AMD just emulates it... this is the main reason why SPi is faster on Intel processors.

If you want to compare single threaded performance between AMD and Intel there are several better and more representative benchmarks that you can use.

I'm starting to bleave those leaked benchmarks are correct...
Computer
(11 items)
 
  
CPUMotherboardGraphicsRAM
Athlon X4 640 with L3 Unlocked - 4ghz Gigabyte GA-78LMT-USB3 GTX 650 8gb (4x2gb) Corsair Vengance 1600-9-9-9-24 
Hard DriveCoolingOSKeyboard
200gb Hitachi Deskstar NH-D14 Win 7 G710+ 
PowerMouseAudio
Corsair TX650v2 G300 Blue Yeti 
  hide details  
Reply
Computer
(11 items)
 
  
CPUMotherboardGraphicsRAM
Athlon X4 640 with L3 Unlocked - 4ghz Gigabyte GA-78LMT-USB3 GTX 650 8gb (4x2gb) Corsair Vengance 1600-9-9-9-24 
Hard DriveCoolingOSKeyboard
200gb Hitachi Deskstar NH-D14 Win 7 G710+ 
PowerMouseAudio
Corsair TX650v2 G300 Blue Yeti 
  hide details  
Reply
post #76 of 116
Quote:
Originally Posted by JCPUser;14681899 
That is NOT the reason. It favors Intel because SPi uses an outdated, rather obsolete instruction set called x87. x87 has basically been replaced by SSE (and it variants) in modern applications. However, Intel still uses x87 on their CPUs wheras AMD just emulates it... this is the main reason why SPi is faster on Intel processors.

If you want to compare single threaded performance between AMD and Intel there are several better and more representative benchmarks that you can use.

Well if that is indeed the case, that makes Bulldozer even better than my napkin math shows.
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
post #77 of 116
Quote:
Originally Posted by kweechy;14681996 
Well if that is indeed the case, that makes Bulldozer even better than my napkin math shows.

Yes : BD is made for gaming (No $h1T!) But it doesnt do heavy number (like sb) crunching. - many things don't require it.
Computer
(11 items)
 
  
CPUMotherboardGraphicsRAM
Athlon X4 640 with L3 Unlocked - 4ghz Gigabyte GA-78LMT-USB3 GTX 650 8gb (4x2gb) Corsair Vengance 1600-9-9-9-24 
Hard DriveCoolingOSKeyboard
200gb Hitachi Deskstar NH-D14 Win 7 G710+ 
PowerMouseAudio
Corsair TX650v2 G300 Blue Yeti 
  hide details  
Reply
Computer
(11 items)
 
  
CPUMotherboardGraphicsRAM
Athlon X4 640 with L3 Unlocked - 4ghz Gigabyte GA-78LMT-USB3 GTX 650 8gb (4x2gb) Corsair Vengance 1600-9-9-9-24 
Hard DriveCoolingOSKeyboard
200gb Hitachi Deskstar NH-D14 Win 7 G710+ 
PowerMouseAudio
Corsair TX650v2 G300 Blue Yeti 
  hide details  
Reply
post #78 of 116
Quote:
Originally Posted by theamdman;14681994 
I'm starting to bleave those leaked benchmarks are correct...

Not sure what my post has to do with leaked benchmarks...confused.gif

I am just explaining one of the reasons AMD is very slow in Spi compared to Intel. It really has nothing to do with BD as the same explanation applies to Phenom II. The only reason AMD beat Intel back in the day with SPi was because Netburst sucked.
post #79 of 116
Quote:
Originally Posted by theamdman;14682023 
Yes : BD is made for gaming (No $h1T!) But it doesnt do heavy number (like sb) crunching. - many things don't require it.

No. Exact opposite.

SB for gaming where the powerful cores reign king on lightly threaded code. BD for rendering where all 8 cores will be 100% 24 hours a day and completely owning SB from the looks of it.
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
2P Workstation
(13 items)
 
  
CPUCPUMotherboardGraphics
Xeon E5 2699 v3 Xeon E5 2699 v3 ASUS Z10PE-D16 EVGA TITAN X 
RAMHard DriveHard DriveCooling
128GB Crucial DDR4 @ 2133MHz Intel 750 Series 1.2TB 4 x 4TB Constellation RAID5 2 x XSPC Raystorms 
CoolingCoolingMonitorMonitor
RS480 MCP655 Dell U3011 Dell U3011 
Case
Silverstone TJ07 
  hide details  
Reply
post #80 of 116
Quote:
Originally Posted by kweechy;14682049 
No. Exact opposite.

SB for gaming where the powerful cores reign king on lightly threaded code. BD for rendering where all 8 cores will be 100% 24 hours a day and completely owning SB from the looks of it.
proof.gif Do you have documentation that permits that what you are saying is worth mentioning?
BLACK!CE
(13 items)
 
  
CPUMotherboardGraphicsRAM
2500k Z68 Exetreme4 Gen3 MSI 6950 Twin Frozr II 8g 8-9-8 24 1600 
Power
CM 800w Silent Pro Gold 
  hide details  
Reply
BLACK!CE
(13 items)
 
  
CPUMotherboardGraphicsRAM
2500k Z68 Exetreme4 Gen3 MSI 6950 Twin Frozr II 8g 8-9-8 24 1600 
Power
CM 800w Silent Pro Gold 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Rumors and Unconfirmed Articles
Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [YT] AMD FX Bulldozer @ 6.5 GHz - under LN2