![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
-
Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.
What is SSE?
- Thread starter Dakna23
- Start date
SSE is Intel's Streaming SIMD Extensions.
It's a set of 8 new 128-bit floating point registers, and 70 new instructions.
Anyways, SSE when worked into some applications can lead to a significant performance increase.
For more information, see
http://www.tommesani.com/SSE.html
It's a set of 8 new 128-bit floating point registers, and 70 new instructions.
Anyways, SSE when worked into some applications can lead to a significant performance increase.
For more information, see
http://www.tommesani.com/SSE.html
peacetilence
2[H]4U
- Joined
- Sep 23, 2004
- Messages
- 2,343
I noticed it says this is Intel based.
I am considering downloading (BELOW see red) to check my AMD 64x2 4400+
Will it work with this processor?
3DMark®05 Free Version (Build 1.2.0)
Full Install Download
Minimum System Recommendations
x86 compatible processor with SSE support, 2GHz
512MB of RAM
1.5GB of free hard disk space
Windows® 2000/XP with the latest updates & Service Packs
DirectX®9.0c or later
DirectX®9.0 compliant graphics adapter that has 128MB of memory and is capable of PixelShader 2.0 or higher
Microsoft® Internet Explorer 6
I am considering downloading (BELOW see red) to check my AMD 64x2 4400+
Will it work with this processor?
3DMark®05 Free Version (Build 1.2.0)
Full Install Download
Minimum System Recommendations
x86 compatible processor with SSE support, 2GHz
512MB of RAM
1.5GB of free hard disk space
Windows® 2000/XP with the latest updates & Service Packs
DirectX®9.0c or later
DirectX®9.0 compliant graphics adapter that has 128MB of memory and is capable of PixelShader 2.0 or higher
Microsoft® Internet Explorer 6
DarkBahamut
Gawd
- Joined
- Mar 18, 2005
- Messages
- 551
Yes, the X2 4400+ has SSE, SSE2 and SSE3
Intel created SSE, but shares the technology with AMD via a cross-licensing agreement. There is no compatibility issues I'm aware of. However, the implementations might be different internally, and due to architectural differences, AMD and Intel processors may benefit from SSE code differently from one another.Dakna23 said:
(cf)Eclipse
Freelance Overclocker
- Joined
- Feb 18, 2003
- Messages
- 30,027
yes, traditionally intel's CPUs see a much larger benefit from SSE than amd'sxonik said:
BillR
Born Again Cynic
- Joined
- Feb 17, 2002
- Messages
- 18,535
(cf)Eclipse said:
The other side of that coin is, AMD CPUs as a rule have much better overall math processors as has been shown time and time again with projects such as distributed computing, such as folding @ home.
I also notice this in Photo Shop and Bryce 3-D.
BillR said:
Thing is, the P4 was designed around SSE2, and the FPU is there only for legacy support. The Athlon64 is based on a design with a fast FPU, and SSE/SSE2 added as an afterthought.
In Windows x64, the FPU will no longer work, so the P4 will finally shake itself from the poor legacy FPU support, and only SSE2 will matter. On the other hand, Athlon64 will lose its legacy FPU advantage. I wonder if this means that with new 64-bit applications, P4 performance will get closer to Athlon64 performance again.
BillR said:
It's a mixed bag, like everything. With some operations the Intels excel, with others the AMDs excel. No processor wins all benchmarks these days.
I will really miss 387 floating-point math... Oh wait. I use linux. 387 will continue to work fine.
Seriously though, 387 has a number of instructions which SSE just doesn't support. FYL2X comes to mind first, there were a couple of other "missing" instructions that I can't recall off the top of my head. Now, it's not that SSE can't compute the same thing as these missing instructions, it just can't do them as fast (software vs. hardware support). The layout (stack-based) of 387 may be a little nasty, but 387 has some really cool instructions, and it's a shame to throw all those away. Perhaps SSE4 will add some of the missing hardware support, but I'm not holding my breath.
Seriously though, 387 has a number of instructions which SSE just doesn't support. FYL2X comes to mind first, there were a couple of other "missing" instructions that I can't recall off the top of my head. Now, it's not that SSE can't compute the same thing as these missing instructions, it just can't do them as fast (software vs. hardware support). The layout (stack-based) of 387 may be a little nasty, but 387 has some really cool instructions, and it's a shame to throw all those away. Perhaps SSE4 will add some of the missing hardware support, but I'm not holding my breath.
visaris said:
Depends on how the CPU works, doesn't it?
If the CPU emulates an FYL2X with a set of SSE2 operations anyway, then there's no point in having it in hardware.
In VS.NET's libraries, the entire C/C++ float library seems to be written in optimized SSE2 code, I assume they did this because it is actually faster on P4. Would be interesting to benchmark sometime perhaps.
(cf)Eclipse
Freelance Overclocker
- Joined
- Feb 18, 2003
- Messages
- 30,027
due to the architecture of because of microsoft's choice? if the former, this surprises me considerably when you think about who developed x86-64Scali said:
(cf)Eclipse said:
It's Microsoft's choice. They simply don't preserve the FPU-context when doing a context-switch. Speeds up multitasking a tad, but as a result, FPU/MMX/3DNow! code will have undefined behaviour.
And well, regardless of who developed it, 90% of all x86-64 CPUs will be Intel.
Well gee, i'd sure love to step into reality -- Could you point me to some sales figures showing intels 64bit parts outselling AMD's? I did some research but all i found was that intel has not released sales figures regarding their itanium, and a lot of information showing they aren't doing very well.
For example:
http://www.theinquirer.net/?article=10668
For example:
http://www.theinquirer.net/?article=10668
Intel still owns over 80% of the x86 market, despite the superiority of the K8 architecture. Even the Opteron has had an uphill battle to 10% of the server market share. When you figure that Intel is also shipping 64bit Celerons and that 6xx P4 are EM64T aware as well, AMD still has a long way to go.
Your comparison of Opteron to Itanium is flawed - Opteron competes with EM64T Xeons, not Itanium. Yes, the Itanic has been a dismal failure, but that's in IA-64, not X86-64.
Your comparison of Opteron to Itanium is flawed - Opteron competes with EM64T Xeons, not Itanium. Yes, the Itanic has been a dismal failure, but that's in IA-64, not X86-64.
Scali said:
True. Both those are valid concerns. Considering that the VS libraries are written for SSE, and the P4 has bad 387 performance, MS dropping 387 support won't matter to many people.
However, for AMD64 cpus running applications compiled with gcc (I've only tested this on linux), and assuming you use the flag "-mfpmath=387" (for some systems this is the default), you can make good use of some of the more nifty 387 instructions. FYL2X in particular can make pretty short work of log() or exp() operations. I've seen some really dramatic differences personally (around 10x as fast), though my code was literally packed full of exp() and log() operations.