StarDestroyer.Net BBS

While it is true that the P4's crippled x87 gets beat handily by K7's FPU, many companies are optimizing their applications for the P4's SIMD core, at which point the advantage swings back to Intel's solution.

I realise that this isn't useful for every application, but I don't find it as cut-and-dry as "AMD FPU superior, P4 x87 inferior!"

Enlightenment wrote:

phongn wrote: Photoshop 6, Word+PowerPoint 2000, Acrobat Distiller, IE6, Mozilla 1.0, AIM, Folding@home, SSH Client, OpenSSH Server, XP Pro IIS. That's on my AXP 1.53GHz.

Again, most of that stuff is going to be spending %99 of its time blocked on network IO or user input. All you're really testing here is if you have enough memory to keep all that stuff in RAM withou thrashing. For a real test of CPU multitasking abilities, run six or seven distributed computing clients (e.g. Folding@Home, SETI@Home, etc, etc, etc) at once.

True indeed, I just felt like posting something

I haven't run multiple dc clients at once (I tried folding & genome, but the two wouldn't play nice with each other). I did have mutlitple processes running, though - like a Photoshop filter running along with Distiller making a PDF. Not like that torture test above, but still somewhat impressive.

It depends on what you're planning to do. For what I do, one ultra-powerful FPU is the most important thing. Snappy response for lots of simultaneous apps is irrelevant; when I hit "execute" and it crunches mutely for two hours, that's a problem. For others, they want the snappy response, and sheer number-crunching power isn't that important.

Darth Wong wrote:It depends on what you're planning to do. For what I do, one ultra-powerful FPU is the most important thing. Snappy response for lots of simultaneous apps is irrelevant; when I hit "execute" and it crunches mutely for two hours, that's a problem. For others, they want the snappy response, and sheer number-crunching power isn't that important.

Oh, I know for engineering and scientific computing the Pentium 4 is a rather poor choice. As I said (and as you note), different solutions for different problems.

Any thoughts about Hammer or PPC970? They're supposed to have very good CPU performance and are both 64bit, which may or may not be of use to what you do.

btw, phongn, it's x86, NOT x87



the 'hammer series of CPUS are capable of handling 64-bit native while still giving full compatibility down to 8 bit, whereas the itanium (assuming this is what you meant by PPG970) is 64-bit native with an already known to be shaky 32-bit compatibility, and nothing below that.

personally i can't wait till the clawhammer and sledgehammer come out.



note to wong: the intel will do the 2 minute sitting with its thumbs up its ass thing if you run LW70 and winamp simultaineously, wheras the AMD K7 will keep on running as if it wasn't lugged out, granted if you try to open a new proggie, you're in for a wait, with either proc.

Hyperion wrote:btw, phongn, it's x86, NOT x87

I am specifically referring to the FPU. It is often referred to as x87, much as IA32 is often referred to as x86.

the 'hammer series of CPUS are capable of handling 64-bit native while still giving full compatibility down to 8 bit, whereas the itanium (assuming this is what you meant by PPG970) is 64-bit native with an already known to be shaky 32-bit compatibility, and nothing below that.

Hammer uses x86-64. Yes, it will run both the x86 instruction set natively and AMD's 64-bit instruction set.

Itanium is not PPC970. The later is a derivative of IBM's POWER4 that is intended for the desktop. It is a 64bit processor that will run 32bit code natively, though the OS will require some modification first. (PPC32 is a subset of PPC64).

phongn wrote:

Hyperion wrote:btw, phongn, it's x86, NOT x87

I am specifically referring to the FPU. It is often referred to as x87, much as IA32 is often referred to as x86.

the 'hammer series of CPUS are capable of handling 64-bit native while still giving full compatibility down to 8 bit, whereas the itanium (assuming this is what you meant by PPG970) is 64-bit native with an already known to be shaky 32-bit compatibility, and nothing below that.

Hammer uses x86-64. Yes, it will run both the x86 instruction set natively and AMD's 64-bit instruction set.

Itanium is not PPC970. The later is a derivative of IBM's POWER4 that is intended for the desktop. It is a 64bit processor that will run 32bit code natively, though the OS will require some modification first. (PPC32 is a subset of PPC64).

ok, thanks for the updated information... though one question, if i'm right in my observation that PPC32 and PPC64 refer to the native bit-level? so on that line wouldn't PPC970 be a 970-bit system? or am i just being dense again?

When I say PPC32 and PPC64, I meant the instruction set. The former is a subset of the latter.

When I say PPC970, PPC7455, PPC750 or other things I'm referring to a processor (GPUL, G4e and G3, respectively).

phongn wrote:When I say PPC32 and PPC64, I meant the instruction set. The former is a subset of the latter.

When I say PPC970, PPC7455, PPC750 or other things I'm referring to a processor (GPUL, G4e and G3, respectively).

ok, thanks. gives me more data.

I perfer my AMD Chips thank you very much.

Of course nothing compares to the nightmare I had with that Cyrix chiped motherboard I bought in '97 and replaced with an AMD in '97, fuck that sucker lasted until I tripped over my sharps container with a coffee mug in my hand. (Reminder never room with a female Diabetic that is messier then you are.)

cyrix sucks, though they are good if you put a frying pan on top of the chip, they actually do work like p-60s in that regard, they can be used for cooking.

I posted a link about the PPC970 earlier. ArsTechnica also has some earlier articles about the earlier revisions of the G4e.

There really is no "PPC32" or "PPC64" instruction set. PowerPC was designed for 64-bit from the beginning, which is why its migration to 64-bit will be a whole fuckload easier than x86.

Acclamator wrote:Discuss

Mine's an AMD Athlon gig. Got it because it was cheaper to build it on this ASUS than an Intel would have been, and I have no gripes.

Durandal wrote:There really is no "PPC32" or "PPC64" instruction set. PowerPC was designed for 64-bit from the beginning, which is why its migration to 64-bit will be a whole fuckload easier than x86.

I was only using PPC32 and PPC64 the clarify my posts. When I meant PPC32 I meant code written for the PPC that is 32bit, as opposed to PPC code that is 64bit.

Ah, right.

Durandal wrote:Ah, right.

Seriously now, did you expect me to make that kind of blatant mistake?

Just keeping you on your toes. :P

Durandal wrote:Just keeping you on your toes.

Bah, I just posted it so I could get a higher post count

phongn wrote:

Durandal wrote:Just keeping you on your toes.

Bah, I just posted it so I could get a higher post count

Well you've succeded beyond your wildest imagination, so what are you gonna do nect? Post a sample instruction set to comapre them and spark a highly technical debate in structure?

CmdrWilkens wrote:

phongn wrote:

Durandal wrote:Just keeping you on your toes.

Bah, I just posted it so I could get a higher post count

Well you've succeded beyond your wildest imagination, so what are you gonna do nect? Post a sample instruction set to comapre them and spark a highly technical debate in structure?

I'm a bio major, fool, not a CS major.

phongn wrote:

CmdrWilkens wrote: Well you've succeded beyond your wildest imagination, so what are you gonna do nect? Post a sample instruction set to comapre them and spark a highly technical debate in structure?

I'm a bio major, fool, not a CS major.

Yes and I'm a History major who used to be an Aerospace major...so what?

And you seriously expect me to compare the instruction sets of IA32 and PowerPC?

phongn wrote:And you seriously expect me to compare the instruction sets of IA32 and PowerPC?

No, I was just stating that it would be the next logical progression unless you want to grab Dalton and start arguing A/V formats

Death to them both! EMP resistant Vacuum tubes forever!!!