Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [WCCF]AMD To Launch Richland APUs on 19th March – Piledriver FX CPU Refresh Expected in June 2013
New Posts  All Forums:Forum Nav:

[WCCF]AMD To Launch Richland APUs on 19th March – Piledriver FX CPU Refresh Expected in June 2013 - Page 6

post #51 of 83
Quote:
Originally Posted by ComputerRestore View Post

Thanks. I always enjoy to ask these questions. I'll go back to my closet now....ph34r-smiley.gif

Yeah, if it was that simple, Bulldozer and Piledriver would fare much better than they are now.

It isn't easy to split a single thread into multiple threads, and most of the time it is downright impossible because the next instruction relies on the results of the previous ones.
Millenium Falcon
(24 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 4930k MSI Big Bang Xpower II EVGA GTX 690 Patriot Viper II Sector 7 
Hard DriveHard DriveHard DriveHard Drive
OCZ Deneva 2 Corsair Force 3 Maxtor Western Digital Green 
Optical DriveCoolingCoolingCooling
Samsung BD/DVD-RW Swiftech MCP655 x2 Black Ice GTX 480 Black Ice GTX 280 
CoolingCoolingCoolingCooling
Alphacool Repack Dual D5 Watercool Heatkiller 3.0 Alphacool GTX 690 fullcover Bitspower Big Bang Xpower II fullcover 
OSMonitorKeyboardPower
Windows 8.1 64-bit Professional 3x Dell S2340 Max Keyboard Durandal CoolerMaster V1000 
CaseMouseMouse PadAudio
Azza Genesis 9000B Logitech G700 Roccat Alumic Onkyo HT-S9100THX 
  hide details  
Reply
Millenium Falcon
(24 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 4930k MSI Big Bang Xpower II EVGA GTX 690 Patriot Viper II Sector 7 
Hard DriveHard DriveHard DriveHard Drive
OCZ Deneva 2 Corsair Force 3 Maxtor Western Digital Green 
Optical DriveCoolingCoolingCooling
Samsung BD/DVD-RW Swiftech MCP655 x2 Black Ice GTX 480 Black Ice GTX 280 
CoolingCoolingCoolingCooling
Alphacool Repack Dual D5 Watercool Heatkiller 3.0 Alphacool GTX 690 fullcover Bitspower Big Bang Xpower II fullcover 
OSMonitorKeyboardPower
Windows 8.1 64-bit Professional 3x Dell S2340 Max Keyboard Durandal CoolerMaster V1000 
CaseMouseMouse PadAudio
Azza Genesis 9000B Logitech G700 Roccat Alumic Onkyo HT-S9100THX 
  hide details  
Reply
post #52 of 83
Quote:
Originally Posted by Tsumi View Post

Yeah, if it was that simple, Bulldozer and Piledriver would fare much better than they are now.

It isn't easy to split a single thread into multiple threads, and most of the time it is downright impossible because the next instruction relies on the results of the previous ones.

Yeah, I can only imagine what would happen if half the instructions on one core were ok, and half on the other got messed up. It'd all be held up by half a single threaded instruction lost in the pipes.

I was half hoping someone more knowledgeable would join in on the theory crafting. bahaha.
Still looking forward to the next revision. smile.gif
Edited by ComputerRestore - 2/7/13 at 1:03pm
post #53 of 83
Quote:
Originally Posted by ComputerRestore View Post

I was thinking more along the lines of - cut the single threaded code in half and divide it between the two cores in a module. So half the work per core. It would be strange to patch it back together at the end, but isn't that how Multithreaded code works?

It is not.

Multithreaded code is a case where a developer very carefully architects a system where a single problem can be solved by processing pieces of the problem independently. If you have a 64x64 pixel bitmap and you want to add 3 blue to each pixel, you can see that it would be relatively simple to divide that bitmap into smaller tiles and have cores claim available tiles when they're idle. When all tiles are processed, the problem is finished and you can enjoy your slightly bluer bitmap in its entirety. The same basic concept holds true for asymmetric problems such as processing audio on one thread and video in another when building a media decoder, but synchronization becomes more complex. The sort of problems found in a game engine become very difficult to split up. Certain pieces are parallelized, and you can see the result of these efforts as over time games have gone from using 1 core to multiple cores more effectively.

The real issue here is that many of the most interesting problems depend on the results of previous stages in order to be able to be processed. Let's take a common problem as an example: add + multiply. The very latest processors include FMA instructions that can do these two things together, but it will serve as a good illustration regardless. If you want to add five to a number and then multiply that result by seven, I think you will readily see a problem. You can't multiply that new number until you've finished adding to it first. There is no way to meaningfully issue that multiply instruction to a second core while the add is still running on the first. You need that result before the multiply makes sense. This is why you can't just rip apart a thread and feed it to multiple cores, and it's also part of why you will not see "reverse hyperthreading".

Note that a single thread in a single core already does get ripped apart and its instructions are executed out of order when possible. The CPU is pretty good at determining dependency over a small range of sequential instructions and it will reorder them to maximize utilization. Instructions are retired in the order they came in, however, so the program executes as expected.
post #54 of 83
I graphed the Overclock potential of each of my FX 8350 Modules (by disabling Individual Modules)



A lot of the FX CPU's have a lot of variance between Core/Module quality (hence how we end up with 4,6,8 Core models)



Is it because of the lengthy pathway from Memory to the L3 Cache? If this was improved upon, could we see faster running CPU's with less heat and voltage?



With a new Socket and the updates to Excavator, maybe we'll end up with something like this ^?
(FX 8670 1.25v 5.2Ghz (Turbo 6.0Ghz) 22nm
Edited by ComputerRestore - 2/8/13 at 4:56pm
post #55 of 83
hope its AM3+
Rog Assassin
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX 8320 Asus Crosshair V Formula 990FX PNY 560ti OC2 G.Skill RipjawsX 1600 CL9 
Hard DriveOSMonitorPower
Crucial M4 64GB SSD Windows 8 Pro x64 VIZIO 42" LCD 1080p Coolermaster GX650 
Case
NZXT Phantom White 
  hide details  
Reply
Rog Assassin
(13 items)
 
  
CPUMotherboardGraphicsRAM
AMD FX 8320 Asus Crosshair V Formula 990FX PNY 560ti OC2 G.Skill RipjawsX 1600 CL9 
Hard DriveOSMonitorPower
Crucial M4 64GB SSD Windows 8 Pro x64 VIZIO 42" LCD 1080p Coolermaster GX650 
Case
NZXT Phantom White 
  hide details  
Reply
post #56 of 83
Quote:
Originally Posted by Rayleyne View Post

Can't people just be happy that new stuff is coming out?
+1
Omega
(17 items)
 
The Beast
(15 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 3930k C2 @ 4.9 Ghz ASUS Rampage IV Formula Quad SLI GTX TITAN 32GB DDR3 2133 Ripjaws Z CAS9 
Hard DriveHard DriveHard DriveHard Drive
256GB x2 Crucial M4 SSD RAID0 Samsung 840 500GB SSD Internal 1TB x 2 External USB 3.0 1.5TB + 2TB + 1TB + 3TB 
Optical DriveCoolingOSMonitor
LG Bluray Burner Triple thick EK 360 Rad with D5 Vario & Raystor... Windows 7 Ultimate x64 Catleap 27" 2560x1440 LED LCD 
KeyboardPowerCaseMouse
Razer Lycosa Coolermaster Silent Pro Gold 1200W Rosewill Thor v2 Logitech G500 
Audio
Logitech 5.1 Surround + Arctic 5.1 Surround Hea... 
CPUMotherboardGraphicsRAM
Intel Core i5 2500k @4.7ghz Asus P8P67 WS Revolution XFX 7970 Black Edition x3 Tri-Fire 8GB DDR3 Gskill 9-9-9-24 
Hard DriveOptical DriveCoolingOS
2TB External+128GB M4 SSD+64GB SSD+250GB WD External DVD Multi H212+ HSF Windows 7 Ultimate 64bit 
MonitorKeyboardPowerCase
Acer GD235HZ 120hz 3d 24" Logitech K120 Corsair TX850 + Thermaltake 850W Antec 900 
MouseAudio
Logitech G500 Logitech 5.1 Surround 
  hide details  
Reply
Omega
(17 items)
 
The Beast
(15 items)
 
 
CPUMotherboardGraphicsRAM
Intel i7 3930k C2 @ 4.9 Ghz ASUS Rampage IV Formula Quad SLI GTX TITAN 32GB DDR3 2133 Ripjaws Z CAS9 
Hard DriveHard DriveHard DriveHard Drive
256GB x2 Crucial M4 SSD RAID0 Samsung 840 500GB SSD Internal 1TB x 2 External USB 3.0 1.5TB + 2TB + 1TB + 3TB 
Optical DriveCoolingOSMonitor
LG Bluray Burner Triple thick EK 360 Rad with D5 Vario & Raystor... Windows 7 Ultimate x64 Catleap 27" 2560x1440 LED LCD 
KeyboardPowerCaseMouse
Razer Lycosa Coolermaster Silent Pro Gold 1200W Rosewill Thor v2 Logitech G500 
Audio
Logitech 5.1 Surround + Arctic 5.1 Surround Hea... 
CPUMotherboardGraphicsRAM
Intel Core i5 2500k @4.7ghz Asus P8P67 WS Revolution XFX 7970 Black Edition x3 Tri-Fire 8GB DDR3 Gskill 9-9-9-24 
Hard DriveOptical DriveCoolingOS
2TB External+128GB M4 SSD+64GB SSD+250GB WD External DVD Multi H212+ HSF Windows 7 Ultimate 64bit 
MonitorKeyboardPowerCase
Acer GD235HZ 120hz 3d 24" Logitech K120 Corsair TX850 + Thermaltake 850W Antec 900 
MouseAudio
Logitech G500 Logitech 5.1 Surround 
  hide details  
Reply
post #57 of 83
I'll be watching closely for this new line to come out, if the am3+ rumor holds true, then i'll buy my brother this cpu and he can buy the motherboard.
post #58 of 83
Quote:
Originally Posted by Particle View Post

Quote:
Originally Posted by ComputerRestore View Post

I was thinking more along the lines of - cut the single threaded code in half and divide it between the two cores in a module. So half the work per core. It would be strange to patch it back together at the end, but isn't that how Multithreaded code works?

It is not.

Multithreaded code is a case where a developer very carefully architects a system where a single problem can be solved by processing pieces of the problem independently. If you have a 64x64 pixel bitmap and you want to add 3 blue to each pixel, you can see that it would be relatively simple to divide that bitmap into smaller tiles and have cores claim available tiles when they're idle. When all tiles are processed, the problem is finished and you can enjoy your slightly bluer bitmap in its entirety. The same basic concept holds true for asymmetric problems such as processing audio on one thread and video in another when building a media decoder, but synchronization becomes more complex. The sort of problems found in a game engine become very difficult to split up. Certain pieces are parallelized, and you can see the result of these efforts as over time games have gone from using 1 core to multiple cores more effectively.

The real issue here is that many of the most interesting problems depend on the results of previous stages in order to be able to be processed. Let's take a common problem as an example: add + multiply. The very latest processors include FMA instructions that can do these two things together, but it will serve as a good illustration regardless. If you want to add five to a number and then multiply that result by seven, I think you will readily see a problem. You can't multiply that new number until you've finished adding to it first. There is no way to meaningfully issue that multiply instruction to a second core while the add is still running on the first. You need that result before the multiply makes sense. This is why you can't just rip apart a thread and feed it to multiple cores, and it's also part of why you will not see "reverse hyperthreading".

Note that a single thread in a single core already does get ripped apart and its instructions are executed out of order when possible. The CPU is pretty good at determining dependency over a small range of sequential instructions and it will reorder them to maximize utilization. Instructions are retired in the order they came in, however, so the program executes as expected.

You're also forgetting stuff that can't be parallelized easily, but can be ran multiple times/need to be ran multiple times, an example being audio encoding...Most of the encoders use 1 thread, but generally good front-ends will detect how many threads you have and convert that many songs at a time, you could do the same with AI for example where each characters AI gets its own thread to run in.
    
CPUMotherboardGraphicsRAM
Intel Core i5 3570k @ 4.5Ghz ASRock Z77 Pro3 Powercolor Radeon HD7950 3GB @ 1150/1350 4x4GB G.Skill Ares 2000Mhz CL9 
Hard DriveHard DriveHard DriveHard Drive
Samsung 840 250GB Western Digital Black 1TB WD1002FAEX Seagate Barracuda 3TB ST3000DM001 Samsung Spinpoint EcoGreen 2TB 
Optical DriveCoolingCoolingCooling
Pioneer DVR-220LBKS Noctua NH-D14 Scythe Gentle Typhoon 1850rpm Corsair AF140 Quiet Edition 
CoolingOSMonitorMonitor
Arcitc Cooling Acclero Twin Turbo II Arch Linux x86-64, amdgpu BenQ G2220HD BenQ G2020HD 
KeyboardPowerCaseMouse
Ducky Shine III Year of the Snake, Cherry Blue Silverstone Strider Plus 600w CoolerMaster CM690 II Black and White SteelSeries Sensei Professional 
Mouse PadAudioOther
Artisan Hien Mid Japan Black Large ASUS Xonar DX NZXT Sentry Mesh 30w Fan Controller 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Intel Core i5 3570k @ 4.5Ghz ASRock Z77 Pro3 Powercolor Radeon HD7950 3GB @ 1150/1350 4x4GB G.Skill Ares 2000Mhz CL9 
Hard DriveHard DriveHard DriveHard Drive
Samsung 840 250GB Western Digital Black 1TB WD1002FAEX Seagate Barracuda 3TB ST3000DM001 Samsung Spinpoint EcoGreen 2TB 
Optical DriveCoolingCoolingCooling
Pioneer DVR-220LBKS Noctua NH-D14 Scythe Gentle Typhoon 1850rpm Corsair AF140 Quiet Edition 
CoolingOSMonitorMonitor
Arcitc Cooling Acclero Twin Turbo II Arch Linux x86-64, amdgpu BenQ G2220HD BenQ G2020HD 
KeyboardPowerCaseMouse
Ducky Shine III Year of the Snake, Cherry Blue Silverstone Strider Plus 600w CoolerMaster CM690 II Black and White SteelSeries Sensei Professional 
Mouse PadAudioOther
Artisan Hien Mid Japan Black Large ASUS Xonar DX NZXT Sentry Mesh 30w Fan Controller 
  hide details  
Reply
post #59 of 83
It's funny how we're really asking for Kaveri and Steamroller to come soon instead of Richland and Piledriver. I definitely won't buy Richland because it's just a "refreshed" version of Trinity even though it has Piledriver cores instead of Bulldozer. I might buy Richland only if the performance is 50%-100% better than Trinity and the price stay dirt cheap. We also realize that AMD will release a greater CPU in the future that is Excavator and that will make us not too interested with their upcoming CPUs before Excavator especially Steamroller. Maybe next time AMD should stop "spreading" their list of upcoming APUs and CPUs. I know that it's good to spread the list so that we know what to expect and plan our budget for the next APUs and CPUs but once we know that they will release a better APUs and CPUs soon, we might not interested with the current and upcoming ones. And that's not good for AMD.
post #60 of 83
Quote:
Originally Posted by Moustache View Post

It's funny how we're really asking for Kaveri and Steamroller to come soon instead of Richland and Piledriver. I definitely won't buy Richland because it's just a "refreshed" version of Trinity even though it has Piledriver cores instead of Bulldozer. I might buy Richland only if the performance is 50%-100% better than Trinity and the price stay dirt cheap. We also realize that AMD will release a greater CPU in the future that is Excavator and that will make us not too interested with their upcoming CPUs before Excavator especially Steamroller. Maybe next time AMD should stop "spreading" their list of upcoming APUs and CPUs. I know that it's good to spread the list so that we know what to expect and plan our budget for the next APUs and CPUs but once we know that they will release a better APUs and CPUs soon, we might not interested with the current and upcoming ones. And that's not good for AMD.

You do realize 50-100% better is virtually impossible with current technology, right? Steamroller at best is only going to bring ~30-40% performance increases, and more likely around 15-20%. The simple clock speed Richland brings is about 5-10%, but it's progress nonetheless.

It doesn't bad that Richland is coming out, if it doesn't affect Kaveri's release schedule. In fact, most of us were expecting Kaveri to come out in 2014, but the new leaks now say in addition to Kaveri being on schedule for 2H 2013, AMD is also releasing Kaveri. Kaveri may be due to AMD working on refining RCM, and that most definitely is a good thing if that's what they're doing. They also aren't releasing when Kaveri will be released, and considering when the next FX CPUs are launching (mid June), it's most likely very late 2013.

Also, Excavator is at best late 2014, if not early 2015. There's no point in talking about tech 2 years from now. Tech a year from now is a stretch at best. The next best thing is always around the corner with computer tech, and if you keep waiting, you'll be waiting forever. If people need a computer now, they're going to buy a computer now because a new CPU coming out in a few months doesn't fulfill their need now.

Not to mention, releasing roadmaps is important for investors. Intel does the same thing. It has been well known that Haswell was launching in mid 2013 since a few months ago.
Millenium Falcon
(24 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 4930k MSI Big Bang Xpower II EVGA GTX 690 Patriot Viper II Sector 7 
Hard DriveHard DriveHard DriveHard Drive
OCZ Deneva 2 Corsair Force 3 Maxtor Western Digital Green 
Optical DriveCoolingCoolingCooling
Samsung BD/DVD-RW Swiftech MCP655 x2 Black Ice GTX 480 Black Ice GTX 280 
CoolingCoolingCoolingCooling
Alphacool Repack Dual D5 Watercool Heatkiller 3.0 Alphacool GTX 690 fullcover Bitspower Big Bang Xpower II fullcover 
OSMonitorKeyboardPower
Windows 8.1 64-bit Professional 3x Dell S2340 Max Keyboard Durandal CoolerMaster V1000 
CaseMouseMouse PadAudio
Azza Genesis 9000B Logitech G700 Roccat Alumic Onkyo HT-S9100THX 
  hide details  
Reply
Millenium Falcon
(24 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 4930k MSI Big Bang Xpower II EVGA GTX 690 Patriot Viper II Sector 7 
Hard DriveHard DriveHard DriveHard Drive
OCZ Deneva 2 Corsair Force 3 Maxtor Western Digital Green 
Optical DriveCoolingCoolingCooling
Samsung BD/DVD-RW Swiftech MCP655 x2 Black Ice GTX 480 Black Ice GTX 280 
CoolingCoolingCoolingCooling
Alphacool Repack Dual D5 Watercool Heatkiller 3.0 Alphacool GTX 690 fullcover Bitspower Big Bang Xpower II fullcover 
OSMonitorKeyboardPower
Windows 8.1 64-bit Professional 3x Dell S2340 Max Keyboard Durandal CoolerMaster V1000 
CaseMouseMouse PadAudio
Azza Genesis 9000B Logitech G700 Roccat Alumic Onkyo HT-S9100THX 
  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 › [WCCF]AMD To Launch Richland APUs on 19th March – Piledriver FX CPU Refresh Expected in June 2013