Overclock.net › Forums › AMD › AMD CPUs › Steamroller?
New Posts  All Forums:Forum Nav:

Steamroller? - Page 327

post #3261 of 3492
Quote:
Originally Posted by Seronx View Post

Full Enabled Category: (bdver4 + bdver5)
SM1+/SP1 Kaveri = <4 (4-way) Cores + <16 GCN CUs + quad-channel DDR4/GDDR5M
SM1+/SP2 Carrizo = <4 (4-way) Cores + <16 GCN CUs + quad-channel DDR4/GDDR5M

Where did you get the SM1+ socket name?
post #3262 of 3492
Quote:
Originally Posted by parvadomus View Post

Where did you get the SM1+ socket name?
It is a guess from the more recent naming conventions that AMD has used before.

FM(x) -> FS(x) -> FP(x)
FT(x) -> FS(x)/AM(x)

SP(x) -> SM(x)
AMD FX ~Seronx
(16 items)
 
  
CPUMotherboardGraphicsRAM
FX-9800P Acer Wasp R7 M440 SK Hynix HMA41GS6AFR8N-TF 
Hard DriveHard DriveOptical DriveCooling
KINGSTON RBU-SNS8152S3128GG2 TOSHIBA MQ01ABD100 HL-DT-ST DVDRAM GUE1N Stock 
OSMonitorKeyboardPower
Microsoft Windows 10 Home Build 14393 Viewsonic XG2401 24 Hz-144 Hz Ducky Channel Shine 3 Stock 65W 
CaseMouseMouse PadAudio
Acer Exoskeleton Steelseries Rival 300 Razer Megasoma AMD-Realtek ALC255 
  hide details  
Reply
AMD FX ~Seronx
(16 items)
 
  
CPUMotherboardGraphicsRAM
FX-9800P Acer Wasp R7 M440 SK Hynix HMA41GS6AFR8N-TF 
Hard DriveHard DriveOptical DriveCooling
KINGSTON RBU-SNS8152S3128GG2 TOSHIBA MQ01ABD100 HL-DT-ST DVDRAM GUE1N Stock 
OSMonitorKeyboardPower
Microsoft Windows 10 Home Build 14393 Viewsonic XG2401 24 Hz-144 Hz Ducky Channel Shine 3 Stock 65W 
CaseMouseMouse PadAudio
Acer Exoskeleton Steelseries Rival 300 Razer Megasoma AMD-Realtek ALC255 
  hide details  
Reply
post #3263 of 3492
I really doubt there are that many things disabled (for 2.4B transistrs 512 shaders are almost its max, at least looking at cape verde which is 1.5B + 2 steamroller modules 800M xtors?). I only hope there is some DDR5 or DDR4 controller sleeping there.
post #3264 of 3492
Quote:
Originally Posted by parvadomus View Post

I really doubt there are that many things disabled (for 2.4B transistrs 512 shaders are almost its max, at least looking at cape verde which is 1.5B + 2 steamroller modules 800M xtors?). I only hope there is some DDR5 or DDR4 controller sleeping there.
Cape verde fully unlocked is 1,5B but this only has 512 isntead of 640 sp's. This only has 8 compared to 16 for cape verde ROP's and then there is the arch difference going to "gcn 1.1"

It is very possible that there is a lot of stuff locked down. and I'm quite sure there is a GDDR5 controller in there just no DDR4 probably.
post #3265 of 3492
Quote:
Originally Posted by maarten12100 View Post

Cape verde fully unlocked is 1,5B but this only has 512 isntead of 640 sp's. This only has 8 compared to 16 for cape verde ROP's and then there is the arch difference going to "gcn 1.1"

It is very possible that there is a lot of stuff locked down. and I'm quite sure there is a GDDR5 controller in there just no DDR4 probably.

It might have 512 SPs instead of 640, but it also has XDMA and Trueaudio plus HSA.
post #3266 of 3492
Quote:
Originally Posted by parvadomus View Post

It might have 512 SPs instead of 640, but it also has XDMA and Trueaudio plus HSA.
Those dsps will take up very little of the actuall transistor count especially XDMA because it replaces the older bridge bus. AMD would not include it in the 290x if it did take up too much die space you know.

I just hope they did indeed lock parts pf the steamroller module since that would mean steamroller with everything enabled would yield the 30% per clock that most of us expected.
post #3267 of 3492
Quote:
Originally Posted by maarten12100 View Post

Those dsps will take up very little of the actuall transistor count especially XDMA because it replaces the older bridge bus. AMD would not include it in the 290x if it did take up too much die space you know.

I just hope they did indeed lock parts pf the steamroller module since that would mean steamroller with everything enabled would yield the 30% per clock that most of us expected.

It was actually 30% per clock over bulldozer that was quoted in that pretty early slide, not piledriver, and they did pretty well for themselves in some loads i think (recent x264 encoders love Haswell and i think Steamroller, compared to their last-gen counterparts)
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
Insert Name Here
(14 items)
 
  
CPUMotherboardGraphicsRAM
6700k Asus Maximus VIII Hero Gigabyte Aorus Xtreme 1080ti Corsair LPX 2x8GB 3200c16 
Hard DriveHard DriveCoolingOS
Old Seagate HDD Samsung 850 EVO Thermalright Silver Arrow SB-E SE Windows 7 Home Premium 64 bit 
MonitorKeyboardPowerCase
Asus PG258Q (240hz + Gsync) WASDKeyboards.com v1 semi custom w/ mx browns, ... Superflower Golden Green HX550 Air540 
MouseMouse Pad
Logitech G Pro Qck+ 
  hide details  
Reply
post #3268 of 3492
Quote:
Originally Posted by Cyro999 View Post

It was actually 30% per clock over bulldozer that was quoted in that pretty early slide, not piledriver, and they did pretty well for themselves in some loads i think (recent x264 encoders love Haswell and i think Steamroller, compared to their last-gen counterparts)
AMD made claims of taking on Piledriver bottlenecks and showed a multitude of ~30% improvements so some of us assumed a big increase. Seeing how broken it actually is that isn't too much to ask for.



That was SteamrollerA though but the increases are against Piledriver not against Bulldozer the comparison they made was against the Piledriver chips in specific server tasks.
post #3269 of 3492
I don't remember them comparing it to Piledriver back then. The Piledriver parts weren't even out when the original 2012 Hot Chips presentation happened. The 30% ops per cycle increase was compared to Bulldozer rather than Piledriver. Piledriver was roughly 7~15% faster than Bulldozer, so it would make sense for the 30% figure to be over Bulldozer since Steamroller was about the same over Piledriver (7~15% on average, 20+% in other cases.)
post #3270 of 3492
Quote:
Originally Posted by NaroonGTX View Post

I don't remember them comparing it to Piledriver back then. The Piledriver parts weren't even out when the original 2012 Hot Chips presentation happened. The 30% ops per cycle increase was compared to Bulldozer rather than Piledriver. Piledriver was roughly 7~15% faster than Bulldozer, so it would make sense for the 30% figure to be over Bulldozer since Steamroller was about the same over Piledriver (7~15% on average, 20+% in other cases.)
The slides are from august and posted at hot chips indeed the release happened after September in October. They are based on Piledriver numbers. Steamroller btw was more but it has lower clocks due to the process it is on in Kaveri.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD CPUs
Overclock.net › Forums › AMD › AMD CPUs › Steamroller?