YOU made the point that and 1800+ will out-run a P4 & I said that this now
isnt the case & we are in the here and now, we havent gone back in time to
when Quantispeed or Willamette architecture were released.
Your jabbering now, the benchmarks prove it's not any more than 2-5% faster
period & no amount of fluffing about SSE or Hardware Prefetching will change
that. Also a 1500XP is SLOWER on almost all accounts than a 1.4 Tbird
explain that one please!!
What, you'd actually WANT your computer to keep running if the CPU's
overheating? That's braindead.
It should of course turn off, to avoid reducing the life expectancy of the
CPU. The clock throttling of the P4 is not a good thing in my opinion as
there's no indication of a hardware problem when the machine's performance
starts to stutter. It looks more like a software issue.
People could run overheating P4s for any length of time and NEVER KNOW ABOUT
IT. If they're ignorant enough they'd just assume that's the way things are
supposed to be (and yes, there are people like that out there, heh :-)) On
the other hand, if your system regularly shuts off, sooner or later you're
FORCED to open the hood!
A far more preferable solution.
FYI in a recent test only 5 out of 18 brand new KT333 board TomsHardware
tested had this implemented, the rest still used the ropey diode under the
socket method for temp control. 5 out of 18 in my book isnt most it's more
like minority, but I'm sure it will imporve.
http://www.tomshardware.com/mainboard/02q2/020509/index.html
Well most boards will give an alarm beep through the system speaker to warn
you that it's all going a bit pear shaped, but then you've got to count on
the user trying to figure out what that annoying noise is. Probably find
that they'd think it's crashed and just reset the thing, hardware monitoring
software should be installed as standard on all full systems & people who
build their own and dont use it deserve to kill their cpu.
take the comment in the context it was intended, which was that the 1800+ and
1.8Ghz P4 at the time Quantispeed was introduced DID NOT LINE UP FOR
PERFORMANCE, thus making claims that the quantispeed rating was based off the
P4's performance rather dubious.
How fast does a Tbird run SSE code?
-JB
Okay, I'll rephrase
"Most new boards that I would go near with a bargepole have this feature"
(that Asus A7S333 is lookin' kinda sweet)
-JB
Yet it still makes as much sense as the AMD-AMD rating, as a 1600XP is not
200mhz faster in reald world applications than a 1.4 Tbird, why cant you
just admit that?
The Tbird doesnt run SSE code as I'm sure your aware, but that still doesnt
change the fact that in games that the 1500XP is slower, than the Tbird 1.4
and that a 1600XP is only 2-5% faster. Why cant you just accept this rather
than try to deflect the issue.
DON'T CARE !!!
> >The Tbird doesnt run SSE code as I'm sure your aware, but that still
doesnt
> >change the fact that in games that the 1500XP is slower, than the Tbird
1.4
> >and that a 1600XP is only 2-5% faster. Why cant you just accept this
rather
> >than try to deflect the issue.
> T-Bird has SSE and 3DNow!. It's SSE2 it doesn't have and the Palomino
> does.
Cheers
Mark
Uh, no. :)
-JB
Nope, the Tbird has 3DNow!, no SSE
Palomino has 3DNow! + SSE
Opteron will have 3dnow! + SSE + SSE2
-JB
On average != in a few benchmarks
You have to include the whole picture, that would include SSE aware apps
screaming ahead of the Tbird's fpu units.. which pushes the average up...
-JB
It's the best but you won't buy it?
well THAT makes a whole lot of sense..
-JB
When averaging performance out, you have to include _all_ code.. this includes
things that use SSE, which will be a HELL of a lot faster on a Pally/Tbred,
and those will push up the average.. helloooooo
"on average the Palomino is 10% faster than Tbird" does NOT mean "The Palomino
is always 10% faster than Tbird"
Unless of course the game is SSE optimised...... (or data prefetch friendly)
yes, instances exist of the Pally being slower than a Tbird, instances also
exist of it screaming out ahead... the 10% figure I quoted was a rough
average......
*sigh*
-JB
Sorry, no. Some of the enhanced 3DNow! instructions that the Athlon
originally added (like precaching and some other stuff) use the exact same
opcodes as the equivalent SSE instructions, but that doesn't mean it has
partial SSE support; it's still 3DNow!. Pre-palomino Athlons have no SSE
support at all, Palomino Athlons have no SSE2 support, only standard SSE.
Just look at the processor caps that for example WCPUID shows for a standard
Athlon and a palomino Athlon and you'll see that for yourself...
