Overclock.net › Forums › Industry News › Hardware News › [xbitlabs] AMD: We Are On Track With Steamroller Micro-Architecture in 2013.
New Posts  All Forums:Forum Nav:

[xbitlabs] AMD: We Are On Track With Steamroller Micro-Architecture in 2013. - Page 63  

post #621 of 718
Anyone know if any new games will support 4+ cores? All I know is that BF4 will come with such high-thread support.
Quote:
Originally Posted by computerparts View Post

Ever heard the famous saying? "If a component measures good and sounds good, it is good. If a component sounds good but measures bad, you’re measuring the wrong thing."

Wut? I thought it was "If the manufacturer doesn't pay you anything, then it's bad. If they do pay you, then it's a 10 out 10."

(J/K)
post #622 of 718
guys really get back on topic
post #623 of 718
L.E.G.I.O.N.
(29 items)
 
S.A.M.P.S.O.N.
(18 items)
 
M.I.C.H.E.L.L.E.
(15 items)
 
CPUMotherboardGraphicsGraphics
AMD FX 8350 Asus Sabretooth 990FX AMD Radeon 7950 Windforce AMD Radeon 7950 Windforce 
RAMHard DriveHard DriveOptical Drive
4x 4gb Corsair Veangance DDR3 2x Seagate 1tb 2x M4 Crucial 64gb Raid 0 on 6gbs 2x LG Supermulti 
CoolingCoolingCoolingCooling
Corsair H100 Liquid Cooling 6x Enermax 120mm Blue LED 4x BitFenix Spectre Pros 120mm Blue LED (courte... 2x Antec Tricools 120mm Blue LED 
CoolingCoolingCoolingOS
2x Scythe Slipstream Slim 120mm 6x ACE High Performance 40mm Blue LED (Emergency Overheat System) 2x stock Corsair H1... Windows 7 64 Bit 
MonitorKeyboardPowerCase
3x 19" 1x17" (Games at 4010x1024) Razer Blackwidow Ultimate Corsair Enthusiast 750w TX Lanboy Air Blue (Maxed Fans) 
MouseMouse PadOtherOther
Razer Naga Razer Vespula 6 CCFLs from ELWirePros.com 60ft EL Wire from ELWirePros.com 
OtherOtherOtherOther
Custom LED Feet from PerformancePCs.com LED Toggle Switches from PerformancePCs.com Custom Built "Rig-nition" courtesy of Digkey 3d Printed Sideplates courtesy of Governor Simcoe 
Other
BitFenix Recon 
CPUMotherboardGraphicsGraphics
965 BE Asus M4N98TD Evo 9500GT 9500GT 
GraphicsRAMHard DriveCooling
5x IO Gear USB-VGA Kingston HyperX WD Blue 500GB Corsair Air Series Quiet Edition 
OSMonitorKeyboardPower
Windows XP (up to) 9x LCD Projectors Generic ToPower 500 
CaseMouseMouse PadAudio
Corsair C70 Green Generic Table Creative Sound Blaster Recon Fatality Professional 
OtherOther
APC 40 APC 20 
CPUMotherboardGraphicsRAM
I7 3770k Gigabyte Z77MITX MSI 7990 Malta "Lightning" Corsair Vengeance 
Hard DriveCoolingCoolingOS
WD Green 4x Spectre Pro 120mm Black H100I Windows 7 Ultimate 
MonitorKeyboardPowerCase
HP 22bw IPS Logitech G510 Corsair CX750 BitFenix Prodigy 
MouseMouse PadAudio
Logitech G700 Ikea Cutting Board Varied 
  hide details  
L.E.G.I.O.N.
(29 items)
 
S.A.M.P.S.O.N.
(18 items)
 
M.I.C.H.E.L.L.E.
(15 items)
 
CPUMotherboardGraphicsGraphics
AMD FX 8350 Asus Sabretooth 990FX AMD Radeon 7950 Windforce AMD Radeon 7950 Windforce 
RAMHard DriveHard DriveOptical Drive
4x 4gb Corsair Veangance DDR3 2x Seagate 1tb 2x M4 Crucial 64gb Raid 0 on 6gbs 2x LG Supermulti 
CoolingCoolingCoolingCooling
Corsair H100 Liquid Cooling 6x Enermax 120mm Blue LED 4x BitFenix Spectre Pros 120mm Blue LED (courte... 2x Antec Tricools 120mm Blue LED 
CoolingCoolingCoolingOS
2x Scythe Slipstream Slim 120mm 6x ACE High Performance 40mm Blue LED (Emergency Overheat System) 2x stock Corsair H1... Windows 7 64 Bit 
MonitorKeyboardPowerCase
3x 19" 1x17" (Games at 4010x1024) Razer Blackwidow Ultimate Corsair Enthusiast 750w TX Lanboy Air Blue (Maxed Fans) 
MouseMouse PadOtherOther
Razer Naga Razer Vespula 6 CCFLs from ELWirePros.com 60ft EL Wire from ELWirePros.com 
OtherOtherOtherOther
Custom LED Feet from PerformancePCs.com LED Toggle Switches from PerformancePCs.com Custom Built "Rig-nition" courtesy of Digkey 3d Printed Sideplates courtesy of Governor Simcoe 
Other
BitFenix Recon 
CPUMotherboardGraphicsGraphics
965 BE Asus M4N98TD Evo 9500GT 9500GT 
GraphicsRAMHard DriveCooling
5x IO Gear USB-VGA Kingston HyperX WD Blue 500GB Corsair Air Series Quiet Edition 
OSMonitorKeyboardPower
Windows XP (up to) 9x LCD Projectors Generic ToPower 500 
CaseMouseMouse PadAudio
Corsair C70 Green Generic Table Creative Sound Blaster Recon Fatality Professional 
OtherOther
APC 40 APC 20 
CPUMotherboardGraphicsRAM
I7 3770k Gigabyte Z77MITX MSI 7990 Malta "Lightning" Corsair Vengeance 
Hard DriveCoolingCoolingOS
WD Green 4x Spectre Pro 120mm Black H100I Windows 7 Ultimate 
MonitorKeyboardPowerCase
HP 22bw IPS Logitech G510 Corsair CX750 BitFenix Prodigy 
MouseMouse PadAudio
Logitech G700 Ikea Cutting Board Varied 
  hide details  
post #624 of 718
Quote:
Originally Posted by A Bad Day View Post

Quote:
Originally Posted by KyadCK View Post

Consoles run Linux, a modified version of windows (Xbox only) or a custom OS.

They can not run .exe files like we do in Windows.

It needs to be recompiled.

Then why not include .exe compatibility so that developers don't have to waste money recompiling?

I'm fairly sure Sony and MS learned the hard way of putting up barricades to porting. Such as the non-standard CPUs.

... Because a recompile is easy. Compile it again.

Adding the ability to run EXEs in other OS's, like WINE, takes years (to build the platform), is unreliable (will this new version of WINE break compatibility with this new program?), and frankly you lose a lot of performance sometimes...

What do you think compiling is? It takes a few hours to compile an entire Linux operating system, games generally have a 20MB EXE.

EDIT: Here's a real life example for you. I play FreeSpace Open. The sourcecode is available to all. The game EXE is ~6.5MB.

It takes my Ph II at 4Ghz 5 minutes to compile the main EXE, the Debug build EXE (20MB), the FRED EXE (mission editor), and the FRED Debug EXE, plus the debug profiles, using MS VS C++.

So why should I add the ability to a custom OS to run an EXE, instead of just recompile?

Oh, also, EXE is a Microsoft thing. Ask Sony why they don't add compatibility.
Quote:
Originally Posted by Ryude View Post

Quote:
Originally Posted by KyadCK View Post

Xbox series, not Xbox the console.
... For someone who knows a lot about linux, you seem to not understand a whole lot about consoles. It requires a recompile. Be it hardware limitation, or a requirement from a development/performance standpoint, it's required.

ummm, yes the Xbox 360 does use a modified NT kernel. If you don't trust me look it up.

I do not understand where this is being misunderstood, so lets try again.

The XBOX SERIES, consisting of the XBOX and XBOX 360 use modified versions of Windows.

What did you misread that made you think I said the Xbox 360 did not run windows?
Quote:
Originally Posted by A Bad Day View Post

Anyone know if any new games will support 4+ cores? All I know is that BF4 will come with such high-thread support.

Since the PS4 will have 8 cores, and the Xbox is rumored or confirmed to have the same, I'm going to say "yes".
Edited by KyadCK - 4/6/13 at 9:34pm
Forge
(17 items)
 
Forge-LT
(7 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5960X (4.625Ghz) ASUS X99-DELUXE/U3.1 EVGA 1080ti SC2 Hybrid EVGA 1080ti SC2 Hybrid 
RAMHard DriveCoolingOS
64GB Corsair Dominator Platinum (3000Mhz 8x8GB) Samsung 950 Pro NVMe 512GB EK Predator 240 Windows 10 Enterprise x64 
MonitorKeyboardPowerCase
2x Acer XR341CK Corsair Vengeance K70 RGB Corsair AX1200 Corsair Graphite 780T 
MouseAudioAudioAudio
Corsair Vengeance M65 RGB Sennheiser HD700 Sound Blaster AE-5 Audio Technica AT4040 
Audio
30ART Mic Tube Amp 
CPUMotherboardGraphicsRAM
i7-4720HQ UX501JW-UB71T GTX 960m 16GB 1600 9-9-9-27 
Hard DriveOSMonitor
512GB PCI-e SSD Windows 10 Pro 4k IPS 
  hide details  
Forge
(17 items)
 
Forge-LT
(7 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5960X (4.625Ghz) ASUS X99-DELUXE/U3.1 EVGA 1080ti SC2 Hybrid EVGA 1080ti SC2 Hybrid 
RAMHard DriveCoolingOS
64GB Corsair Dominator Platinum (3000Mhz 8x8GB) Samsung 950 Pro NVMe 512GB EK Predator 240 Windows 10 Enterprise x64 
MonitorKeyboardPowerCase
2x Acer XR341CK Corsair Vengeance K70 RGB Corsair AX1200 Corsair Graphite 780T 
MouseAudioAudioAudio
Corsair Vengeance M65 RGB Sennheiser HD700 Sound Blaster AE-5 Audio Technica AT4040 
Audio
30ART Mic Tube Amp 
CPUMotherboardGraphicsRAM
i7-4720HQ UX501JW-UB71T GTX 960m 16GB 1600 9-9-9-27 
Hard DriveOSMonitor
512GB PCI-e SSD Windows 10 Pro 4k IPS 
  hide details  
post #625 of 718
Quote:
Originally Posted by KyadCK View Post

... Because a recompile is easy. Compile it again.

Adding the ability to run EXEs in other OS's, like WINE, takes years (to build the platform), is unreliable (will this new version of WINE break compatibility with this new program?), and frankly you lose a lot of performance sometimes...

What do you think compiling is? It takes a few hours to compile an entire Linux operating system, games generally have a 20MB EXE.
I do not understand where this is being misunderstood, so lets try again.

The XBOX SERIES, consisting of the XBOX and XBOX 360 use modified versions of Windows.

What did you misread that made you think I said the Xbox 360 did not run windows?
Since the PS4 will have 8 cores, and the Xbox is rumored or confirmed to have the same, I'm going to say "yes".

That's misleading, a lot. It can take an hour just to compile and link Chrome on some systems, an exe of under 2mb (In linux, it's about 35 min on my rig, maybe 45). The problem is, it takes a long time to compile all the additional content in the application. The exe, or binary executable (for linux nerds), is only a small portion of what is compiled. chrome as pure source is somewhere around 2G of source, that takes a while and this is a considerably small program when you look at the full complexity. An opsys? It doesn't take hours to compile linux, it takes DAYS to compile a full desktop. I repeat, days. A good game could take a single day alone to compile.

Best part? We haven't even talked about how long it takes to get code compiled on a different platform. You have to tweak the code, build it and go back to the build errors. Just getting it to build could take extreme amounts of time and effort. Then you have to get it to build appropriately with the same "performance". So now we are talking about months of work, just to get that console game on windows (or any other platform).

That is why "porting" is so expensive, it isn't as easy as "slap in the code here". It's easier when the hardware is similar but they are going to have some stuff that will need to be re-designed. Direct hardware functions that may or may not be on the new target platform have to be re-written into formats for the new API. That "to the metal" has to be re-done, completely.

In ways, creating a wine like wrapper is going to be easier and better for them. Your "will a new version break things" isn't exactly too big of a deal. Since the hardware is already known, well designed, the only issue would be the software side of things. If Sony/MS designed a wrapper to release games with, the company has to do no work in compiling and re-working code to "port" the game. MS/Sony can keep the wrapper up to date and working with their system. Wine has issues because DX software varies so much, on top of that they have to deal with OGL and constant linux tweaks/updates. Not to mention, the big anchor, they don't know the windows kernel calls, or some of the more in depth DX API functions. They are doing this blind, where a wrapper done by the console creator wouldn't be. It's much easier, development goes faster, performance enhancements come quicker. (Assuming Win plays ball with Sony, a certain amount of money should probably aid in that if it were ever addressed, but either way this is all speculation that parties involved get along which usually doesn't happen)

The goal of a wrapper isn't to emulate anything but to make the opsys communicate with the game as if it communicates naturally. In theory the overhead of performance should be minimal, the problem is knowing both sides of the equation so you can balance it equally (an apt comparison).

I'm not defending or saying they should, just that you aren't really making a clear picture of this. Porting is a TON of work alone, especially when you spend so much time tweaking your code to run on the metal and then scrapping that because it won't compile on the new target platform. Essentially ports don't exist per-se, what happens is features just stop being added and the code just gets thrown into the compile phase. Meaning no new development on the engine or anything, it runs as is once they get it to compile and build. So you get the same game, or as close to it as can be done. It sucks because you pay the same price for either but one runs on a cheaper solution.
Edited by mushroomboy - 4/6/13 at 9:49pm
Current Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 4.6GHz@1.44v GA-990FXA-UD3 R4.0 HD 7950 (1100/1450) 8G Muskin DDR3 1866@8CLS 
Hard DriveOptical DriveOSMonitor
1TB WD LiteOn DVD-RW DL Linux/Windows 19" Phillips TV 1080p 
PowerCaseMouseMouse Pad
OCZ 600W Generic Junk Logitech MX400 Generic Junk 
Audio
SBL 5.1 
  hide details  
Current Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 4.6GHz@1.44v GA-990FXA-UD3 R4.0 HD 7950 (1100/1450) 8G Muskin DDR3 1866@8CLS 
Hard DriveOptical DriveOSMonitor
1TB WD LiteOn DVD-RW DL Linux/Windows 19" Phillips TV 1080p 
PowerCaseMouseMouse Pad
OCZ 600W Generic Junk Logitech MX400 Generic Junk 
Audio
SBL 5.1 
  hide details  
post #626 of 718
Quote:
Originally Posted by mushroomboy View Post

Quote:
Originally Posted by KyadCK View Post

... Because a recompile is easy. Compile it again.

Adding the ability to run EXEs in other OS's, like WINE, takes years (to build the platform), is unreliable (will this new version of WINE break compatibility with this new program?), and frankly you lose a lot of performance sometimes...

What do you think compiling is? It takes a few hours to compile an entire Linux operating system, games generally have a 20MB EXE.
I do not understand where this is being misunderstood, so lets try again.

The XBOX SERIES, consisting of the XBOX and XBOX 360 use modified versions of Windows.

What did you misread that made you think I said the Xbox 360 did not run windows?
Since the PS4 will have 8 cores, and the Xbox is rumored or confirmed to have the same, I'm going to say "yes".

Long Post

I wrote out an example of a real life cross platform game (x86/OpenGL) which is what the new console games will be closer to. That's more than enough to prove my point. When you have real numbers in front of you to prove a game compile time, I'll hear it. I'll go get the FSO source code again so I won't be a hypocrite if you like too. They have guides on how to compile on linux distros as well, so you can try it too. Yes I know this wasn't in your quote, so you may have missed it.

This "perfect world wrapper" is crap, and will never happen. You know that too. MS and Sony directly compete. Sony won't share bluray, MS won't share DirectX (or Sony chooses not to pay). Let alone working on a wrapper together.

Not going to argue with the point of a wrapper, because that was never the goal. Only that it's a bad idea to use one. Recompile is easy, they're on the same arch. You, as a linux user, know this. It's rewriting to use OpenGL that'll be hard.


As a side note, I would like to point out that your compile times are extremely far off.



An hour huh? No one who compiles for a living (like a AAA game developer) uses such slow CPUs. Just like to point that out.
Edited by KyadCK - 4/6/13 at 10:06pm
Forge
(17 items)
 
Forge-LT
(7 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5960X (4.625Ghz) ASUS X99-DELUXE/U3.1 EVGA 1080ti SC2 Hybrid EVGA 1080ti SC2 Hybrid 
RAMHard DriveCoolingOS
64GB Corsair Dominator Platinum (3000Mhz 8x8GB) Samsung 950 Pro NVMe 512GB EK Predator 240 Windows 10 Enterprise x64 
MonitorKeyboardPowerCase
2x Acer XR341CK Corsair Vengeance K70 RGB Corsair AX1200 Corsair Graphite 780T 
MouseAudioAudioAudio
Corsair Vengeance M65 RGB Sennheiser HD700 Sound Blaster AE-5 Audio Technica AT4040 
Audio
30ART Mic Tube Amp 
CPUMotherboardGraphicsRAM
i7-4720HQ UX501JW-UB71T GTX 960m 16GB 1600 9-9-9-27 
Hard DriveOSMonitor
512GB PCI-e SSD Windows 10 Pro 4k IPS 
  hide details  
Forge
(17 items)
 
Forge-LT
(7 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5960X (4.625Ghz) ASUS X99-DELUXE/U3.1 EVGA 1080ti SC2 Hybrid EVGA 1080ti SC2 Hybrid 
RAMHard DriveCoolingOS
64GB Corsair Dominator Platinum (3000Mhz 8x8GB) Samsung 950 Pro NVMe 512GB EK Predator 240 Windows 10 Enterprise x64 
MonitorKeyboardPowerCase
2x Acer XR341CK Corsair Vengeance K70 RGB Corsair AX1200 Corsair Graphite 780T 
MouseAudioAudioAudio
Corsair Vengeance M65 RGB Sennheiser HD700 Sound Blaster AE-5 Audio Technica AT4040 
Audio
30ART Mic Tube Amp 
CPUMotherboardGraphicsRAM
i7-4720HQ UX501JW-UB71T GTX 960m 16GB 1600 9-9-9-27 
Hard DriveOSMonitor
512GB PCI-e SSD Windows 10 Pro 4k IPS 
  hide details  
post #627 of 718
Quote:
Originally Posted by KyadCK View Post

I wrote out an example of a real life cross platform game (x86/OpenGL) which is what the new console games will be closer to. That's more than enough to prove my point. When you have real numbers in front of you to prove a game compile time, I'll hear it. I'll go get the FSO source code again so I won't be a hypocrite if you like too. They have guides on how to compile on linux distros as well, so you can try it too. Yes I know this wasn't in your quote, so you may have missed it.

This "perfect world wrapper" is crap, and will never happen. You know that too. MS and Sony directly compete. Sony won't share bluray, MS won't share DirectX (or Sony chooses not to pay). Let alone working on a wrapper together.

Not going to argue with the point of a wrapper, because that was never the goal. Only that it's a bad idea to use one. Recompile is easy, they're on the same arch. You, as a linux user, know this. It's rewriting to use OpenGL that'll be hard.


As a side note, I would like to point out that your compile times are extremely far off.

-=snip=-

An hour huh? No one who compiles for a living (like a AAA game developer) uses such slow CPUs. Just like to point that out.

Compile times depend on flags, libs used, compiler, linker, it all plays into effect. My times weren't that off, I gave 35 min on my system with 8GB ram. Where's that chart says a Phenom II x4 falls? About 30 min, which is actually off too. Linux is known to do faster compiles than Windows, don't remember why. So if they are getting a quicker compile time, they have some flags disabled that I'm running. I can assure you of that. I also didn't say it was definite, it can take up to an hour. I gave a rough estimate of what my system gets, 35 min, maybe 45. Now I was guessing because I didn't remember, hence the two numbers. It probably took me a lil' over 30 minutes, maybe longer. As I've said variables play into this.

Note I'm not wrong about Gentoo, it will take days to get a stage 3 going to a desktop. And no, I won't accept the bare system. I'm talking full opsys, it's going to take days. Were talking all the essentials, all the user space tools, working up to a fully fledged desktop with something like KDE. That takes quite a while, building KDE (last I tried) took a decent amount of time. The biggest problem about building source isn't the code, it's linking. A game is assuredly going to take a good amount of time, I'd say easily a day. [edit] Holy crap, they've now done parallelization for gentoo compiles. Well ain't that a sour one. Guess Gentoo isn't such a pain now, well that takes my entire stance off. I was trying to find times, found times alright but apparently they've changed things since I last went at it. Well, regardless of that, building a single program as large as a game is going to take a while. A good while. Note, I'm not saying it takes the entire day, 24hrs, but it's going to take a full work day to just compile the code. During the initial build stages it's going to be hell, especially if the code doesn't want to build. That's what is really going to take them the longest. Cause you simply don't just scour code, it's easier to build a project and let make fail then spit out the error.

And we can't exactly say how the porting process will work, new gen isn't out. They already stated the chips are going to be custom, the desktop variant of the PS4 is going to be "crippled" for whatever that means. If they code to the metal, it's going to have to be re-done anyways even if it is on x86. Even porting a game from Windows to Linux, with OGL, creates a ton of work. What you are forgetting is, this isn't exactly X86. Their are enough major differences from this to Intel that would make porting to a PC platform a good nuance. Can't exactly port to "amd only" users? That's a little bias, so whatever amd specific functions (AVX) are going to have to be removed. It's not going to be nearly as easy as going from Windows to Linux, it will be more work. Not as much work going from PPC to X86, that is for sure.
Edited by mushroomboy - 4/6/13 at 10:52pm
Current Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 4.6GHz@1.44v GA-990FXA-UD3 R4.0 HD 7950 (1100/1450) 8G Muskin DDR3 1866@8CLS 
Hard DriveOptical DriveOSMonitor
1TB WD LiteOn DVD-RW DL Linux/Windows 19" Phillips TV 1080p 
PowerCaseMouseMouse Pad
OCZ 600W Generic Junk Logitech MX400 Generic Junk 
Audio
SBL 5.1 
  hide details  
Current Rig
(14 items)
 
  
CPUMotherboardGraphicsRAM
FX-8350 4.6GHz@1.44v GA-990FXA-UD3 R4.0 HD 7950 (1100/1450) 8G Muskin DDR3 1866@8CLS 
Hard DriveOptical DriveOSMonitor
1TB WD LiteOn DVD-RW DL Linux/Windows 19" Phillips TV 1080p 
PowerCaseMouseMouse Pad
OCZ 600W Generic Junk Logitech MX400 Generic Junk 
Audio
SBL 5.1 
  hide details  
post #628 of 718
Quote:
Originally Posted by Stay Puft View Post

Don't start with the intel conspiracy crap in this thread.

Do you mean it's 'crap' that Intel threatens their own customers (which is proven beyond question already), or do you mean to suggest that the fact that Intel threatens its own customers is somehow untrue, despite mountains of evidence and sword affidavits from employees of several of the major OEM manufacturers who can't possibly have any vested interest in making this stuff up?

If the former, I agree with you. It is 'crappy' that Intel threatens customers and pays off companies for years (like Dell) not to use AMD chips.

If the later, you really need to start reading a little more. You can start with this New York Times article here:
http://www.nytimes.com/2009/11/05/technology/companies/05chip.html?_r=0
Edited by anubis44 - 4/7/13 at 12:09am
Current Rig
(12 items)
 
  
CPUMotherboardGraphicsRAM
AMD R7 1700 Gigabyte AX370-Gaming 5 GIgabyte Windforce R9 290 G.Skill Trident Z 
Hard DriveOptical DriveCoolingOS
Crucial MX100 Samsung DVD AMD Ryzen stock HSF Win 10 x64 
MonitorKeyboardPowerCase
22" 1680x1050 Samsung  1989 IBM Model M PS/2 'clicky' keyboard (best k... EVGA 850w Cooler Master Storm Stryker 
  hide details  
Current Rig
(12 items)
 
  
CPUMotherboardGraphicsRAM
AMD R7 1700 Gigabyte AX370-Gaming 5 GIgabyte Windforce R9 290 G.Skill Trident Z 
Hard DriveOptical DriveCoolingOS
Crucial MX100 Samsung DVD AMD Ryzen stock HSF Win 10 x64 
MonitorKeyboardPowerCase
22" 1680x1050 Samsung  1989 IBM Model M PS/2 'clicky' keyboard (best k... EVGA 850w Cooler Master Storm Stryker 
  hide details  
post #629 of 718
Quote:
Originally Posted by Artikbot View Post

Could be. And most certainly it is easier than reducing the pipeline, considering there's only been one year between the release of Piledriver and Steamroller.

7 stage pipeline is the ideal on paper... As you said in reality it would mean way too low clocks to be competitive with deeper processors at much higher speeds. Mind you, I wouldn't care the slightest if in increasing the IPC about this said 30%, the processor's max clock under a hefty water loop fell to 4.5, or even 4.4GHz. Heck, even to Phenom II levels. We would still have eight cores with an IPC a good 20% faster than Phenom II.

Now that you comment on the micro-op cache. AMD said something about that, let me dig it up!

Edit: There we go. Steamroller DOES feature a micro op queue. Perhaps the misprediction improvement comes from here?

The issue is, even the Athlon64 had a 12 stage pipeline and struggeld to hit over 3Ghz, obviously clocks do improve over time but I have a feeling a 7 stage one would struggle..otherwise you can bet Intel and AMD would have done it.

I'd wager a lot of the improvement does, but also from improved prediction...Maybe the ideal one would be say, 12 stages with a microop cache, that way it could be the equivalent of a 7 stage pipeline when it can skip the decoding stage but still hit some decent clock speeds assuming you can skip ~5 stages. (Just pulling numbers out of the air, so obviously take it with a grain of salt)
Quote:
Originally Posted by Artikbot View Post

Nothing to be surprised of, most Windows software used (I don't know for sure now, but most certainly still is) Intel's x86 compiler, which pretty obviously, didn't have any kind of support for AMD-exclusive flags.

It also uses (used?) a round-a-bout method of working out if a CPU has flags. (Rather than detect the flags themselves, it uses/at least it used the family flags, which means an AMD CPU that supports say, SSE4.1 and SSE4.2 still won't use it. See here for proof.)
Quote:
Originally Posted by 2010rig View Post

I guess that's great for 5% of computer users out there. Check this out, Linux has been hovering around the 3% - 5% mark for 8 years.

You should know web results vary dramatically and aren't really that accurate, even Steve Ballmer thought Linux has over-taken OS X while some sites are saying under 1% still...And I'd imagine a great portion of Ubuntu users won't ever go on w3schools, as it is I know that 95% of the time I'm on it, it's when I'm at Uni/Work and on a PC without my OS, etc choices on it. (That said, I still program in Windows usually)
Quote:
Originally Posted by Ryude View Post

Look at the fact that Intel decides when it changes process nodes, while AMD has to wait on GF or TSMC.

Not only that, but Intel has long had the best fabs in the industry by far. The last time AMD (Or anyone) really got ahead was when AMD beat Intel to the 130nm node, or when their switch to the 180nm node was filled with a lot less problems than the Intel switch.
Quote:
Originally Posted by 2010rig View Post

AMD ALREADY settled those allegations, and the FTC ALREADY took care of it, what more do you want?


Looking at the Linux vs Windows differences here, it's pretty clear Intel is still doing it at a much lower level than they were, an FX-8350 goes from competing fairly well with an i7 3770k in Linux to competing with an i5 3570k in Windows...Most likely AMD just hasn't got the money to actually fight it still.
Quote:
Originally Posted by HanSomPa View Post

Anand Bench CPU comparison.
http://www.anandtech.com/bench/Product/701?vs=697

i5 3570K
http://www.newegg.com/Product/Product.aspx?Item=N82E16819116504
FX 8350
http://www.newegg.com/Product/Product.aspx?Item=N82E16819113284

i5 wins in like 80% of the benchmarks? On top of using less power, unless you have a very specific workload, why should you buy an FX8350? You can also OC the i5 more than the FX I think. Not sure, haven't looked around at OC.net's average OCs yet.

Okay, with these benchmarks you have to remember to do some stuff to adjust it for the real world:
1) Take out Sysmark, we have plenty of real world/B] benchmarks/programs in there, why use a synthetic?
2) Take out the single-core results (ie. Cinebench 1 core) because while interesting to look at per-core performance, are you going to run CAD rendering, etc on a single core? Niche use at best.
3) Use the FX-8320, it's closer in clock speed to the i5 so you get true IPC results.

AMD is either a bit behind, or ahead (7zip even beating an i7) depending on the program until you get gaming results.

Quote:
Originally Posted by looniam View Post

edit: oh yeah, how is the PCI 3.0 working out on those . .oh nevermind . . .

The equivalent bandwidth is equal due to AMD having more lanes than nearly every Z77 board. (Bar the ultra expensive ones with a PCIe bridge)
    
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  
    
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  
post #630 of 718
Quote:
Originally Posted by Brutuz View Post


I'd wager a lot of the improvement does, but also from improved prediction...Maybe the ideal one would be say, 12 stages with a microop cache, that way it could be the equivalent of a 7 stage pipeline when it can skip the decoding stage but still hit some decent clock speeds assuming you can skip ~5 stages. (Just pulling numbers out of the air, so obviously take it with a grain of salt)
It also uses (used?) a round-a-bout method of working out if a CPU has flags. (Rather than detect the flags themselves, it uses/at least it used the family flags, which means an AMD CPU that supports say, SSE4.1 and SSE4.2 still won't use it. See here for proof.)
Pretty bad Microsoft should push out a compiler which is equal for both platforms this is just bad.
We need better windows competition and this seems like a easy way.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
This thread is locked  
Overclock.net › Forums › Industry News › Hardware News › [xbitlabs] AMD: We Are On Track With Steamroller Micro-Architecture in 2013.