Overclock.net › Forums › AMD › AMD CPUs › Is AMD going to x86 instead of staying x64? Why?
New Posts  All Forums:Forum Nav:

Is AMD going to x86 instead of staying x64? Why? - Page 4

post #31 of 74
Not to sound offensive, but how can you list yourself as a "programmer" and ask this question?

--Another programmer
SG06
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5-3570S 65W ASRock Z77E-ITX VisionTek AMD Radeon RX 480 8GB Reference G.SKILL Sniper DDR3-2400 (@ 1866 9-10-10-28) 2x8GB 
Hard DriveHard DriveCoolingCooling
Crucial m4 256GB Toshiba P300 3TB Prolimatech Samuel 17 Noctua NF-F12 
OSMonitorMonitorKeyboard
Windows 10 Pro x64 HP Omen 32" HP Pavilion 27xi Das Model S Professional Silent 
PowerCaseMouseAudio
Silverstone SST-ST45SF Silverstone SG06 Microsoft Intellimouse Realtek ALC898 
  hide details  
Reply
SG06
(16 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i5-3570S 65W ASRock Z77E-ITX VisionTek AMD Radeon RX 480 8GB Reference G.SKILL Sniper DDR3-2400 (@ 1866 9-10-10-28) 2x8GB 
Hard DriveHard DriveCoolingCooling
Crucial m4 256GB Toshiba P300 3TB Prolimatech Samuel 17 Noctua NF-F12 
OSMonitorMonitorKeyboard
Windows 10 Pro x64 HP Omen 32" HP Pavilion 27xi Das Model S Professional Silent 
PowerCaseMouseAudio
Silverstone SST-ST45SF Silverstone SG06 Microsoft Intellimouse Realtek ALC898 
  hide details  
Reply
post #32 of 74
Wikipedia is a great resource, but it is still prone to misinformation, typos, and vandalism. If something doesn't seem right, or if you don't know enough about the topic to know if it sounds right or not, make sure you check other sources.

Quote:
Originally Posted by nathris View Post
Actually you can. Intel's AVX is a 256 bit instruction set, running on a 64bit processor. You can have 256 bit registers alongside 64 bit ones.
Well known, already happening (we've been using 128-bit instructions since the P3), and not what he was discounting.

I do not believe it is practical to run an x86 OS just in the AVX (or any other purpose built SIMD or MIMD) registers. The general purpose registers would have to be 128-bit as well, or face extreme overhead penalties/slowdowns when trying to emulate 128-bit.
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.3/3.6GHz core/uncore, 1.225/1.2v Gigabyte X99 SOC Champion (F22n) 2x Sapphire R9 290X Tri-X OC New Edition (10036... 4x4GiB Crucial @ 2667, 12-11-12-27-T1, 1.37v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.05) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Cooler Master Nepton 280L 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Reference R9 290X w/Stilt's MLU 1000e / 1375m E... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
Primary
(15 items)
 
Secondary
(13 items)
 
In progress
(10 items)
 
CPUMotherboardGraphicsRAM
5820K @ 4.3/3.6GHz core/uncore, 1.225/1.2v Gigabyte X99 SOC Champion (F22n) 2x Sapphire R9 290X Tri-X OC New Edition (10036... 4x4GiB Crucial @ 2667, 12-11-12-27-T1, 1.37v 
Hard DriveHard DriveHard DriveCooling
Plextor M6e 128GB (fw 1.05) M.2 (PCI-E 2.0 2x) 2x Crucial M4 256GB 4x WD Scorpio Black 500GB Cooler Master Nepton 280L 
OSMonitorKeyboardPower
Windows 7 Professional x64 SP1 BenQ BL3200PT Filco Majestouch Tenkeyless (MX Brown) Corsair RM1000x 
CaseMouseAudio
Fractal Design Define R4 Logitech G402 Realtek ALC1150 + M-Audio AV40 
CPUMotherboardGraphicsRAM
X5670 @ 4.4/3.2GHz core/uncore, 1.36 vcore, 1.2... Gigabyte X58A-UD5 r2.0 w/FF3mod10 BIOS Reference R9 290X w/Stilt's MLU 1000e / 1375m E... 2x Samsung MV-3V4G3D/US @ 2000, 10-11-11-30-T1,... 
RAMHard DriveHard DriveHard Drive
1x Crucial BLT4G3D1608ET3LX0 @ 2000, 10-11-11-3... OCZ (Toshiba) Trion 150 120GB Hyundai Sapphire 120GB 3x Hitachi Deskstar 7k1000.C 1TB 
CoolingOSPowerCase
Noctua NH-D14 Windows 7 Pro x64 SP1 Antec TP-750 Fractal Design R5 
Audio
ASUS Xonar DS 
CPUMotherboardGraphicsRAM
i7-6800K @ 4.3/3.5GHz core/uncore, 1.36/1.2v ASRock X99 OC Formula (P3.10) GTX 780 (temporary) 4x4GiB Crucial DDR4-2400 @ 11-13-12-28-T2, 1.33v 
Hard DriveHard DriveCoolingOS
Intel 600p 256GB NVMe 2x HGST Travelstar 7k1000 1TB Corsair H55 (temporary) Windows Server 2016 Datacenter 
PowerCase
Seasonic SS-860XP2 Corsair Carbide Air 540 
  hide details  
Reply
post #33 of 74
Quote:
Originally Posted by BallaTheFeared View Post
That said floating point on a cpu is nearly dead, which is one of the reasons AMD cut theirs in half.

128 bit per core, capable of combing into one 256 bit pipeline, but where the SB chips will have 8 256bit pipelines on their 8 core cpu, bulldozer will only have 4 on its 8 core cpu (edit: 4 256 bit pipelines combined or 8 126 bit pipelines)
OK, so this is a lie. AMD did NOT cut thier FPU in half.

Both Intel and AMD have 128-bit FP, we have an FMAC, they have an FPU.

To run a 256-bit AVX instruction AMD combines two FMACs and Intel combines their 128-bit FPU and their SSE registers.

So, AMD can run an AVX and an SSE execution at the same time, Intel cannot.

Nobody has a pure 256-bit AVX unit. Based on how much FPU happens in most applications that would be a massive waste of die space and create a lot of power draw/heat.
post #34 of 74
Quote:
Originally Posted by JF-AMD View Post
OK, so this is a lie. AMD did NOT cut thier FPU in half.

Both Intel and AMD have 128-bit FP, we have an FMAC, they have an FPU.

To run a 256-bit AVX instruction AMD combines two FMACs and Intel combines their 128-bit FPU and their SSE registers.

So, AMD can run an AVX and an SSE execution at the same time, Intel cannot.

Nobody has a pure 256-bit AVX unit. Based on how much FPU happens in most applications that would be a massive waste of die space and create a lot of power draw/heat.
that makes efficiency worse, i am right ?


for example, this guy should gain score closer to 153 GFLOPS, instead of 123 ...
http://www.overclock.net/intel-cpus/...l#post12223248


what i found, is k10 (thuban) have 6-7% less GFLOPS score in linx or IBT (than their theoretical maximum)

Sandybridge have around 25% drop (AVX), 10-11% drop in nonAVX mode, from intel known performance numbers

Edited by pietro sk - 2/18/11 at 5:55am
    
CPUMotherboardGraphicsRAM
Opteron 6386SE G34r1 H8SGL Radeon R9 380X 32GB ECC_DDR3 1333 4ch 
Hard DriveHard DriveCoolingOS
Seagate 600 Pro Enterprise SSD WD Green+Black some water Win8.1_64 
KeyboardPowerCaseMouse
K120 TX850 Tai-chi M560 
AudioAudio
Xmos_u8_DAC; (diy Pluto 2A3/SE estat amp) STAX cookies 3pcs 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Opteron 6386SE G34r1 H8SGL Radeon R9 380X 32GB ECC_DDR3 1333 4ch 
Hard DriveHard DriveCoolingOS
Seagate 600 Pro Enterprise SSD WD Green+Black some water Win8.1_64 
KeyboardPowerCaseMouse
K120 TX850 Tai-chi M560 
AudioAudio
Xmos_u8_DAC; (diy Pluto 2A3/SE estat amp) STAX cookies 3pcs 
  hide details  
Reply
post #35 of 74
Looking at a 16-core Interlagos vs. an 8-core Sandybridge in the server space:

256-bit throughput should be about the same
AMD should have 2X the 128-bit throughput (16 FMACs vs. 8 FPUs.)
For 256-bit with lots of SSE, AMD should do better because AMD can do a 256-bit AVX and an SSE on the same cycle. Intel cannot do both and would have to run them on 2 cycles. Intel recommends taking 128-bit SSE instructions and recoding them as AVX-128; but AVX-128 pads the upper register above 128-bit, so even though you have a 256-bit AVX path, you can only run either a 128-bit OR a 256-bit on any cycle. You can't run 2 128-bit AVX through a 256-bit as I understand it (from the presentation at IDF, page 8, SF10_ARCS004_100.pdf).
post #36 of 74
Thread Starter 
Something leads me to believe that there is some unknown (or severely unpopular) technology some engineers know about which is one reason why the chip makers are shifting to CPU/GPU....

It leads me to believe that possibly Parallel &/or DNA processing or some other bizarre conceptual processing design is about to be developed or come to developement which will revolutionize &/or significantly increase a CPU's Processing Performance....

Even though that could be a remote possibility, we have other bottlenecks out there, like the HDD/SSD hardware, which still really isn't up to part with something on the lines of a PCIe SSD & still not fast enough to push enough data to these super light year processing beast we have & are about to come into our hands (Bulldozer)...

Then one looks at the server side where multiple CPUs are on one Main Board (I like this word more than Motherboard), which could have been developed in the background by AMD Engineers to test new theorems & concepts..

Call me crazy, but I hope we see some significant advances into the Super Computer department some time in the near future... It's not so much about which platform you compute on, but the possibility of what a desktop can do....

(Transcoding large Video files faster, Compressing Archives, and Compiling Large Code faster would be what I'm referring to here.. Also some new Open GL stuff too..)
The Rock
(15 items)
 
  
Reply
The Rock
(15 items)
 
  
Reply
post #37 of 74
Quote:
Originally Posted by JF-AMD View Post
OK, so this is a lie. AMD did NOT cut thier FPU in half.

Both Intel and AMD have 128-bit FP, we have an FMAC, they have an FPU.

To run a 256-bit AVX instruction AMD combines two FMACs and Intel combines their 128-bit FPU and their SSE registers.

So, AMD can run an AVX and an SSE execution at the same time, Intel cannot.

Nobody has a pure 256-bit AVX unit. Based on how much FPU happens in most applications that would be a massive waste of die space and create a lot of power draw/heat.
Its not a lie, I was just simply wrong.

Like AMDs marketing guy was lying when he said Phenom I was going to be 40% faster than C2Q?

I still fail to see why floating point even matters on the home user end.
    
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 #38 of 74
Thread Starter 
Quote:
Originally Posted by thrasherht View Post
+Rep, thanks for the source...
The Rock
(15 items)
 
  
Reply
The Rock
(15 items)
 
  
Reply
post #39 of 74
Thread Starter 
Quote:
Originally Posted by Chranny View Post
The reason I posted FAIL (which is one of the very few post I have done this in) was because the entire concept of the article stemmed upon speculation & data which could never be gathered/obtained, therefore the entire thread was irrelevant, and furthermore any possible conclusion at best wouldn't even be remotely accurate...

I wasn't disrespecting anyone, I was saying the article/source & reasoning was a fail...
The Rock
(15 items)
 
  
Reply
The Rock
(15 items)
 
  
Reply
post #40 of 74
Quote:
Originally Posted by BallaTheFeared View Post
I still fail to see why floating point even matters on the home user end.
intelburntest / linx bragging rights, perhaps?
    
CPUMotherboardGraphicsRAM
Opteron 6386SE G34r1 H8SGL Radeon R9 380X 32GB ECC_DDR3 1333 4ch 
Hard DriveHard DriveCoolingOS
Seagate 600 Pro Enterprise SSD WD Green+Black some water Win8.1_64 
KeyboardPowerCaseMouse
K120 TX850 Tai-chi M560 
AudioAudio
Xmos_u8_DAC; (diy Pluto 2A3/SE estat amp) STAX cookies 3pcs 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Opteron 6386SE G34r1 H8SGL Radeon R9 380X 32GB ECC_DDR3 1333 4ch 
Hard DriveHard DriveCoolingOS
Seagate 600 Pro Enterprise SSD WD Green+Black some water Win8.1_64 
KeyboardPowerCaseMouse
K120 TX850 Tai-chi M560 
AudioAudio
Xmos_u8_DAC; (diy Pluto 2A3/SE estat amp) STAX cookies 3pcs 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD CPUs
Overclock.net › Forums › AMD › AMD CPUs › Is AMD going to x86 instead of staying x64? Why?