[quote] Obviously to me, unless I'm mistakin the 1.6GHz Power PC 7455 @ 1600 is out and they're cranking away. My source is from distributed.net. Here is the page where they post their speeds for the rc5 projects. The info most likely should be under NDA, but woops. Scroll down the list of PPC until ...Bingo! <hr></blockquote>
<strong>Uh, you should probably look <a href="http://e-www.motorola.com/webapp/sps/site/taxonomy.jsp?nodeId=03M943030450467M98653" target="_blank">here.</a> There's no such thing as a 1600MHz 7455. There's no such things as 733, 800, 867, 933, or 1066MHz 7400's, either. There's something seriously wrong with that list...</strong><hr></blockquote>
But there is a 2 x 800, a 2 x 533, and getting the 7400/7450 distinction mixed up wouldn't surprise me at all. No surprises here, folks.
[quote]Uh, you should probably look here. There's no such thing as a 1600MHz 7455. There's no such things as 733, 800, 867, 933, or 1066MHz 7400's, either. There's something seriously wrong with that list... <hr></blockquote>
Someone want to tell me what in the heck this guy is talking about?
Sorry for being a dick....I'm just totally lost as to the point of your post...and 2x 1000 Mhz not equal 2000 Mhz, btw.
[ 06-26-2002: Message edited by: The All Knowing 1 ]</strong><hr></blockquote>
I would have previously agreed that 2x 1000 MHz does not equal 2000 MHz, and frankly I don't see how it could in reality but then I saw this on distrubited.net:
[quote]... the keyrate of any individual processor within a multi-processor machine will generally be comparable to that of identical MHz single-processor machine. As such, the overall speed for a multiprocessor machine is generally merely the speed of one processor multiplied by the number of processors within that machine. <hr></blockquote>
Here is the basis for my assertion that dual 1.2 w/ DDR would be "substantially" faster than the current 1.0 ghz duals (the simple version being that the dual 1ghz Xserve is "substantially" faster than the dual 1ghz Powermac in these benchmarks.
[quote] But there is a 2 x 800, a 2 x 533, and getting the 7400/7450 distinction mixed up wouldn't surprise me at all. No surprises here, folks. <hr></blockquote>
Right, but there's no dual Gig Mac (2000) on the list. I find it kinda difficult to believe that no one with a dual Gig machine is interested in distributed.net.
[edit-- then again, there's only one person with a dual 800 interested, so who knows.]
What all this means is that the criteria for entries in the list is poorly defined. It really shouldn't be used as an accurate benchmark.
... the keyrate of any individual processor within a multi-processor machine will generally be comparable to that of identical MHz single-processor machine. As such, the overall speed for a multiprocessor machine is generally merely the speed of one processor multiplied by the number of processors within that machine.
perhaps someone can put this in perspective. <hr></blockquote>
I'd imagine it's because the code + data for RC5 is small enough to fit entirely within the cache of a modern processor. Since the processes wouldn't have to frequently access the bus, there's little chance of them bumping into each other for access to memory/disk.
But there is a 2 x 800, a 2 x 533, and getting the 7400/7450 distinction mixed up wouldn't surprise me at all. No surprises here, folks.</strong><hr></blockquote>
well it doesn't seem to be a case of a simple mistake, distributed.net has a seperate page for multi-processor speeds
Could these be over-clocked "standard issue" processors? No, I don't think you could overclock a 1Ghz to 1.6 GHz? The site itself doesn't seem to be bogus, but they rely on info from users. Could someone be giving them bad info, and they have yet to catch it? You can probably tell this is really bugging me- huh?
I believe the Distributed.net benchmarks page lists USER-REPORTED cpu data. In other words, if you report your CPU to them and say it's a G5 2.4ghz, then I guess that's what they list on their chart.
It seems to me, guys, you mix up several incompatible, or, rather, irrelevant notions here.
1) MHz paranoia. Surely (and it's already a proved fact) performance is not only CPU clock speed. In this sense Apple is much less lagging behind, if at all, than most people on earth think.
2) Mhz paranoia. Since even a 3-yo child knows that 1 GHz is less than 3 GHz (as most people on earth believe), WIntel's sales are doing well. They inspire the MHz rush by making you think that your 1GHz box is out-of-date just because Windows 2003 will require at least a 3GHz box. Yeah, from the sales point of view, Apple has a more difficult task to convince users that Steve's RDF is closer to reality than Intel's. Because in the Wintel world MHz rise faster.
So, guys, your FUD is justified in the sense of marketshare, profit and Apple's commercial future, while it's obvious to me that in the sense of the future of computing and technology all is not lost for Apple.
Perhaps, it's one of the reasons why Apple feeds us the recent Switch ads.
I was more amazed at the quad 604e. Ok aathat could theoretically exist in non Apple hardware and so could the others but the 1.6... Oh hell, I give up -back to reality I just wanna believe .
Once again, if anyone can shed some light on what these tests are, etc.
Thanks
<hr></blockquote>
Dslreports RC5 team here - the speed listings are submitted by users. The "1600" guy just submitted his dual 800 as 1600 - there is no 1600 CPU. He added the 2 800's in the dual 800 together, as advised by the instructions on the page.
The number is correct, though (16 million keys/sec). My dual gHz QuickSilver gets about 19 million keys per second.
Follow the instructions. You can also run it from the Terminal in X with the "-bench" flag, e.g.
./dnetc -bench
and it will do its own benchmark. If your processor is a G4, it will absolutely WASTE any x86 processor, as the Mac client has hand-coded highly optimized Altivec code.
Just to point this out... the quad 604e is real. I don't remember exactly who the company that made those was... all I remember is it was during the Clone Era and I have an old MacWorld on my shelf with a full-page ad for the guys. Talk about a kickin machine for it's time...
<strong>if the dual thing is true...if...what if its because the new iron was delayed by 3 months (i think moki said he heard a rumor that the ddr mobo slipped</strong><hr></blockquote>
I said that a specific DDR mobo had slipped (again), something I heard from a few people I know -- and that you're likely to see another less ambitious DDR mobo. Topping out at 1.2ghz looks likely, but with a 166mhz bus and DDR.
In terms of real world performance, an improved memory subsytem -- especially given OS X's propensity for shuffling around huge amounts of data -- will make a huge difference.
Comments
[ 06-27-2002: Message edited by: Gamblor ]</p>
[quote] Obviously to me, unless I'm mistakin the 1.6GHz Power PC 7455 @ 1600 is out and they're cranking away. My source is from distributed.net. Here is the page where they post their speeds for the rc5 projects. The info most likely should be under NDA, but woops. Scroll down the list of PPC until ...Bingo! <hr></blockquote>
Uh, you should probably look <a href="http://e-www.motorola.com/webapp/sps/site/taxonomy.jsp?nodeId=03M943030450467M98653" target="_blank">here.</a> There's no such thing as a 1600MHz 7455. There's no such things as 733, 800, 867, 933, or 1066MHz 7400's, either. There's something seriously wrong with that list...
<strong>Uh, you should probably look <a href="http://e-www.motorola.com/webapp/sps/site/taxonomy.jsp?nodeId=03M943030450467M98653" target="_blank">here.</a> There's no such thing as a 1600MHz 7455. There's no such things as 733, 800, 867, 933, or 1066MHz 7400's, either. There's something seriously wrong with that list...</strong><hr></blockquote>
But there is a 2 x 800, a 2 x 533, and getting the 7400/7450 distinction mixed up wouldn't surprise me at all. No surprises here, folks.
We will see soon enough.
<strong>
Someone want to tell me what in the heck this guy is talking about?
Sorry for being a dick....I'm just totally lost as to the point of your post...and 2x 1000 Mhz not equal 2000 Mhz, btw.
[ 06-26-2002: Message edited by: The All Knowing 1 ]</strong><hr></blockquote>
I would have previously agreed that 2x 1000 MHz does not equal 2000 MHz, and frankly I don't see how it could in reality but then I saw this on distrubited.net:
[quote]... the keyrate of any individual processor within a multi-processor machine will generally be comparable to that of identical MHz single-processor machine. As such, the overall speed for a multiprocessor machine is generally merely the speed of one processor multiplied by the number of processors within that machine. <hr></blockquote>
perhaps someone can put this in perspective.
[ 06-27-2002: Message edited by: pey/coy-ote ]</p>
<strong>
<a href="http://www.xinet.com/benchmarks/benchmarks.2002/index.html" target="_blank">Not necessarily...</a></strong><hr></blockquote>
Nice find, Gamblor. Interesting results, too.
<strong>
Here is the basis for my assertion that dual 1.2 w/ DDR would be "substantially" faster than the current 1.0 ghz duals (the simple version being that the dual 1ghz Xserve is "substantially" faster than the dual 1ghz Powermac in these benchmarks.
<a href="http://www.xinet.com/benchmarks/benchmarks.2002/index.html" target="_blank">http://www.xinet.com/benchmarks/benchmarks.2002/index.html</a></strong><hr></blockquote>
Note that these are benchmarks of I/O intensive operations, at which the Xserve will be clearly superiour to the (current) PowerMac.
Right, but there's no dual Gig Mac (2000) on the list. I find it kinda difficult to believe that no one with a dual Gig machine is interested in distributed.net.
[edit-- then again, there's only one person with a dual 800 interested, so who knows.]
What all this means is that the criteria for entries in the list is poorly defined. It really shouldn't be used as an accurate benchmark.
[quote] quote:
--------------------------------------------------------------------------------
... the keyrate of any individual processor within a multi-processor machine will generally be comparable to that of identical MHz single-processor machine. As such, the overall speed for a multiprocessor machine is generally merely the speed of one processor multiplied by the number of processors within that machine.
--------------------------------------------------------------------------------
perhaps someone can put this in perspective. <hr></blockquote>
I'd imagine it's because the code + data for RC5 is small enough to fit entirely within the cache of a modern processor. Since the processes wouldn't have to frequently access the bus, there's little chance of them bumping into each other for access to memory/disk.
[ 06-27-2002: Message edited by: Gamblor ]</p>
<strong>
But there is a 2 x 800, a 2 x 533, and getting the 7400/7450 distinction mixed up wouldn't surprise me at all. No surprises here, folks.</strong><hr></blockquote>
well it doesn't seem to be a case of a simple mistake, distributed.net has a seperate page for multi-processor speeds
[quote] Power PC 7450/7455 G4\t1333\t(2xcpu)\t rc5\t 27,864,728\t <hr></blockquote><a href="http://n0cgi.distributed.net/speed/query.cgi?cputype=all&arch=all&contest=all&multi=1 " target="_blank">multi-processor speed page</a>
Could these be over-clocked "standard issue" processors? No, I don't think you could overclock a 1Ghz to 1.6 GHz? The site itself doesn't seem to be bogus, but they rely on info from users. Could someone be giving them bad info, and they have yet to catch it? You can probably tell this is really bugging me- huh?
[ 06-27-2002: Message edited by: pey/coy-ote ]</p>
<strong>sizzle chest: Nope. Beat you by two minutes.
</strong><hr></blockquote>
Oops, I saw your 2nd post but not the earlier one!
[quote]Originally posted by Gamblor:
<strong>
Uh, you should probably look <a href="http://e-www.motorola.com/webapp/sps/site/taxonomy.jsp?nodeId=03M943030450467M98653" target="_blank">here.</a> There's no such thing as a 1600MHz 7455. There's no such things as 733, 800, 867, 933, or 1066MHz 7400's, either. There's something seriously wrong with that list... </strong><hr></blockquote>
I believe the Distributed.net benchmarks page lists USER-REPORTED cpu data. In other words, if you report your CPU to them and say it's a G5 2.4ghz, then I guess that's what they list on their chart.
It seems to me, guys, you mix up several incompatible, or, rather, irrelevant notions here.
1) MHz paranoia. Surely (and it's already a proved fact) performance is not only CPU clock speed. In this sense Apple is much less lagging behind, if at all, than most people on earth think.
2) Mhz paranoia. Since even a 3-yo child knows that 1 GHz is less than 3 GHz (as most people on earth believe), WIntel's sales are doing well. They inspire the MHz rush by making you think that your 1GHz box is out-of-date just because Windows 2003 will require at least a 3GHz box. Yeah, from the sales point of view, Apple has a more difficult task to convince users that Steve's RDF is closer to reality than Intel's. Because in the Wintel world MHz rise faster.
So, guys, your FUD is justified in the sense of marketshare, profit and Apple's commercial future, while it's obvious to me that in the sense of the future of computing and technology all is not lost for Apple.
Perhaps, it's one of the reasons why Apple feeds us the recent Switch ads.
I thought we'd all come to the conclusion that the OS & the motherboard are more deserving of attention than a c. 100mhz speed shift?
Upgrading the motherboard significantly will make a FAR bigger difference than simply speeding up the cpu.
The OS will be tweaked to near-perfection and mhz won't be an issue.
Don't blame Apple for the Mhz cock-up anyway - blame Motorola.
Once again, if anyone can shed some light on what these tests are, etc.
Thanks
<hr></blockquote>
Dslreports RC5 team here - the speed listings are submitted by users. The "1600" guy just submitted his dual 800 as 1600 - there is no 1600 CPU. He added the 2 800's in the dual 800 together, as advised by the instructions on the page.
The number is correct, though (16 million keys/sec). My dual gHz QuickSilver gets about 19 million keys per second.
You can download the client from <a href="http://www.distributed.net/download/clients.html" target="_blank">http://www.distributed.net/download/clients.html</a>
Follow the instructions. You can also run it from the Terminal in X with the "-bench" flag, e.g.
./dnetc -bench
and it will do its own benchmark. If your processor is a G4, it will absolutely WASTE any x86 processor, as the Mac client has hand-coded highly optimized Altivec code.
<strong>
Wow. That was a great post. I just can't understand why people give you a hard time. Really, I just don't get it.
<img src="graemlins/oyvey.gif" border="0" alt="[No]" /> <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> </strong><hr></blockquote>
LOL, it's so easy to offend all the geeks here. BTW, I happen to like Adams, I've read most of his books.
<img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
<strong>if the dual thing is true...if...what if its because the new iron was delayed by 3 months (i think moki said he heard a rumor that the ddr mobo slipped</strong><hr></blockquote>
I said that a specific DDR mobo had slipped (again), something I heard from a few people I know -- and that you're likely to see another less ambitious DDR mobo. Topping out at 1.2ghz looks likely, but with a 166mhz bus and DDR.
In terms of real world performance, an improved memory subsytem -- especially given OS X's propensity for shuffling around huge amounts of data -- will make a huge difference.