Yeah i know - that bus is only 50 Mhz slower than my comp.
The 850 Chipset uses dual channel Rambus for theoretical bandwidth of 3.2 GB/sec. But u need 2 Rambus sticks to run that
ARG! that'll cost 800 bucks - almsots as much as the p4 chip.
The 850 Chipset uses dual channel Rambus for theoretical bandwidth of 3.2 GB/sec. But u need 2 Rambus sticks to run that
ARG! that'll cost 800 bucks - almsots as much as the p4 chip.any possibilities of running a P4 w/o rambus?
I like the idea of 1.7 ghz, and 400 FSB, but Rambus? yuk
I like the idea of 1.7 ghz, and 400 FSB, but Rambus? yuk
I hope Intel supports Non-Rambus as well
But Intel isn't licensing the P7 bus architecture right now, so we can't count on VIA or Micron to make a DDR chipset. This could change though.
The Pentium 4 Xeon (aka foster), it is said, will use only use chipsets that support SDRAMS and DDRSDRAMS. If Rambus bombs out, Intel can probably modify them for desktop use.
Personally, I'm waiting to see if this 400 MHz fsb gives rambus the awesome performance boost intel says it will. If Rdram fails here, it is dead forever and I'll happily say good riddance to bad RDRUBBISH!
But Intel isn't licensing the P7 bus architecture right now, so we can't count on VIA or Micron to make a DDR chipset. This could change though.
The Pentium 4 Xeon (aka foster), it is said, will use only use chipsets that support SDRAMS and DDRSDRAMS. If Rambus bombs out, Intel can probably modify them for desktop use.
Personally, I'm waiting to see if this 400 MHz fsb gives rambus the awesome performance boost intel says it will. If Rdram fails here, it is dead forever and I'll happily say good riddance to bad RDRUBBISH!
I highly doubt that GHz would make a big difference in GFLOPs. First off MACs at 500-600 mhz have more MFLOPS than any intel or AMD chip, so you are getting higher performance at lower cpu. its like running a a small motor with more horsepower than a big motor with less hp.
AMD should've caught onto this with their Athlon cpus and really take a bite out of Intel. Too bad that Anand or any other major computer reviewer put a head to head test with a Mac and compare the results. It was the same thing with SPARC that they ran at 300mhz but produced more Mflops than pcs at the time.
So just loading up on MHZ doesn't change the fact that you can have a G4 500-600mhz or whatever they are at beating a Coppermine 1ghz at loading stuff. I guess people like bigger numbers when it comes to Mhz, like Cyrix's PR rating hehe
AMD should've caught onto this with their Athlon cpus and really take a bite out of Intel. Too bad that Anand or any other major computer reviewer put a head to head test with a Mac and compare the results. It was the same thing with SPARC that they ran at 300mhz but produced more Mflops than pcs at the time.
So just loading up on MHZ doesn't change the fact that you can have a G4 500-600mhz or whatever they are at beating a Coppermine 1ghz at loading stuff. I guess people like bigger numbers when it comes to Mhz, like Cyrix's PR rating hehe
Gunman I'd have to disagree with you a bit there..
I beleive that "Loading Stuff" Probably has more to do with hard drive speed than processor. A ! Gig P!!! would certainly lose to a G4 if the Gig had a crummy HDD and the G4 a 10000 RPM SCSI and vice versa.
I have yet to see a MAC with a 600 MHz G4. I know IBM has PwrPCs that run at 700+ Mhz, but these are not the Altivec ones that Motorola makes. The Mac only outperforms Athlon/P!!! when it comes to Altivec enabled applications. I've used G4s, and they are damn fine machines, but I really can't tell the difference btwn a G4 and an x86 in everyday apps (Perhaps if i did more photoshp stuff it would matter).
Games - Never played games on a G4, but I played games on an iMac before and the performance was anemic at best. I've heard the G4 is infinitly better with games.
Yes, the SunMircro UltraSparc II outperforms x86 in FPU by 2 - 2.5x per clock cycle, but its integer performance is no better.
So its really a toss up as to which platform is better. Depends on your OS prefrence really.
I beleive that "Loading Stuff" Probably has more to do with hard drive speed than processor. A ! Gig P!!! would certainly lose to a G4 if the Gig had a crummy HDD and the G4 a 10000 RPM SCSI and vice versa.
I have yet to see a MAC with a 600 MHz G4. I know IBM has PwrPCs that run at 700+ Mhz, but these are not the Altivec ones that Motorola makes. The Mac only outperforms Athlon/P!!! when it comes to Altivec enabled applications. I've used G4s, and they are damn fine machines, but I really can't tell the difference btwn a G4 and an x86 in everyday apps (Perhaps if i did more photoshp stuff it would matter).
Games - Never played games on a G4, but I played games on an iMac before and the performance was anemic at best. I've heard the G4 is infinitly better with games.
Yes, the SunMircro UltraSparc II outperforms x86 in FPU by 2 - 2.5x per clock cycle, but its integer performance is no better.
So its really a toss up as to which platform is better. Depends on your OS prefrence really.
You are also forgeting that just because the CPU is faster clock for clock, can the CPU be ramped up in speed. Look at the Cyrix they coudl never get their CPU's at high speed. Then look at Intel with the P6 core, started at 150 Mhz now soon 1.13 Mhz
1.7 Ghz P4 upon release? I think not.
Lets just assume for a moment that Intel can perform such a feat (quite an assumption IMHO). Intel would NEVER NEVER release a processor 50% faster than the competition. That would be shootin their entire wad in one sitting! A more likely situation would be the advertised 1.3 and 1.4Ghz release forcing AMD to raise their speed as well. Once AMD answered, Intel then releases another processor.....etc. etc. etc. This way, Intel gets the stock ralley every announcement and the big surge in publicity.
On the theory side, I don't think Intel can make a 1.7Ghz processor on a .18um process. Furthermore, Intel is a year or more away from their .13um copper process, so that won't be comming into play any time soon either.
All this sound like BS to me.....but time will tell. I certianly wouldn't mind Intel raising the performance bar again! God isn't competition great?
Lets just assume for a moment that Intel can perform such a feat (quite an assumption IMHO). Intel would NEVER NEVER release a processor 50% faster than the competition. That would be shootin their entire wad in one sitting! A more likely situation would be the advertised 1.3 and 1.4Ghz release forcing AMD to raise their speed as well. Once AMD answered, Intel then releases another processor.....etc. etc. etc. This way, Intel gets the stock ralley every announcement and the big surge in publicity.
On the theory side, I don't think Intel can make a 1.7Ghz processor on a .18um process. Furthermore, Intel is a year or more away from their .13um copper process, so that won't be comming into play any time soon either.
All this sound like BS to me.....but time will tell. I certianly wouldn't mind Intel raising the performance bar again! God isn't competition great?
with the 400Mhz bus...
i think this will be a problem when overclocking, as it isn't a true 400Mhz bus but a 200Mhz DDR or 100Mhz QDR, and if the EV6 bus (100Mhz DDR used on Athlon/duron motherboards) is anything to go by i susspect that it will NOT overclock well at all!
i think this will be a problem when overclocking, as it isn't a true 400Mhz bus but a 200Mhz DDR or 100Mhz QDR, and if the EV6 bus (100Mhz DDR used on Athlon/duron motherboards) is anything to go by i susspect that it will NOT overclock well at all!
You are probably very right about the 200 DDR/400 bus (or whatever it is) not overclocking very well. BUT at 1.4-1.7GHZ I won't be complaining.... woohoo can't wait!
One reason Intel WOULD release a chip that is 50% faster than the compitition: The compitition has been pissing on their shoes lately! Intel HAS to get the momentum back. The PIII architecture couldn't do it for them, but it really looks like the PIV will be able to do the job. I think I will get one when I can pick up a 1.5Gig one for under $600. That damned RDRAM is going to REALLY suck! Hopefully, by the time my processor is $600, the RAM will only be $150 per 128Meg...
Yeah, and the unfortunate truth of the whole matter is that once these super processers are released they will most likely end up performing only marginally better than CPU's several hundred megahertz slower and hundreds of dollars cheaper; thus another letdown, giving us something else to bitch about for months to come.
Cosmic_Horror,
I suspect your overclocking guess is correct. This is most likely the same reason the EV6 does not overclock well also. The clock rate doubling circuit effectively halves the tolerance for the overclocking. I would emagine that another doubler would make the issue even more timing constrained.
According to Dean Kents industry update:
http://www.realworldtech.com/page.cfm?ArticleID=RWT071000000000
The P4 isn't going to be anything more than talk and samples in this calendar year (although it may well be more prominent than Intels pre-announced 1.13 Ghz PIII!).
The inferior FPU of this processor really makes me wonder about Intel's logic. P4 seems to be solely designed to clock to high frequencies even at the expense of performance. Could marketecture be more important in todays CPU market than architecture?
I suspect your overclocking guess is correct. This is most likely the same reason the EV6 does not overclock well also. The clock rate doubling circuit effectively halves the tolerance for the overclocking. I would emagine that another doubler would make the issue even more timing constrained.
According to Dean Kents industry update:
http://www.realworldtech.com/page.cfm?ArticleID=RWT071000000000
The P4 isn't going to be anything more than talk and samples in this calendar year (although it may well be more prominent than Intels pre-announced 1.13 Ghz PIII!).
The inferior FPU of this processor really makes me wonder about Intel's logic. P4 seems to be solely designed to clock to high frequencies even at the expense of performance. Could marketecture be more important in todays CPU market than architecture?
Perhaps raising the FSB woudn't be a good idea with the QDR, but I'm sure someone will come up with a way to modify the chip to defeat the multiplyer lock. Of course, I'm not sure many people will want to risk screwing up an 800 dollar chip.
Concerning the x87 fpu performance - yes, I think the x87 performance will suck if everything I've read is true. But anything written/compiled for sse and sse2 should really fly.
As a last note - this chip will be huge - 170 mm square. Get out those Golden Orbs!
Concerning the x87 fpu performance - yes, I think the x87 performance will suck if everything I've read is true. But anything written/compiled for sse and sse2 should really fly.
As a last note - this chip will be huge - 170 mm square. Get out those Golden Orbs!
I will almost bet anything that the P4 1700 won't hit the streets this year by the looks of things. Remember when they promised the Gigger a long time ago. They launched it in Feb and by may it was still an impossible chip to find. Right now, I will take what intel says and add at least 4 months to it. So they say Sept for a 1.3 gigger, I would not expect it until jan of next year. That would mean the 1.7 won't hit the streets until at least march of next year.
Good luck intel. I need another round of laughter.
Good luck intel. I need another round of laughter.
The P4 uses a 100 MHz QDR bus.
And VIA will be producing DDR chipsets for the P4, intel has licenced them to do that, since they want an option for Rambus, now that everyone cept for the late fkloster hates it, and they're legally forbidden to produce DDR chipsets themselves.
And VIA will be producing DDR chipsets for the P4, intel has licenced them to do that, since they want an option for Rambus, now that everyone cept for the late fkloster hates it, and they're legally forbidden to produce DDR chipsets themselves.
Yeah, Intel has now made a name for itself.... announcing faster chips but not having them out for the public.... they need to work on that and get the chips out faster... oh well dual 700s at 933 or 1gighz will hold me over
The Willy is shaping up to be the inverse of the K5. Remember that chip?? Couldn't ramp worth a damn, but could do a hell of a lot per cycle. Has to do with the number of pipelines, I think. Willy has 20 (if I remember correctly). Why so many?? So they can ramp the hell out of it to look good. For instance, VIA is gonna (maybe) release the Samuel (Joshua, Cayenne, who knows anymore) core at 1.0 GHz...we all know it performs like a Celeron 400, but some schmuck at Best Buy or CompUSA won't. They only see the number of GHz and the most wins. That is why Intel is doing this with the Willy. It won't necessarily perform that well (hell, I'll take a 1.2 GHz Athlon over a 1.4 GHz Willy any day), but it will sell because it will be able to ramp really well. So, for those of us who actually look at performance numbers, the MHz (GHz) number doesn't matter. What matters is how many FPS the chip can pump out.
I still remember merely a year ago, when everyone's got the 400-600 MHZ CPUs and were all making big fuzz about it.... I have a 450MHZ and was very satisfied with it, and look at our new babies here.... I kind of want to save money for now and upgrade may be 2 or 3 years later, but damn.. I wonder how fast the computers are gonna be two or three years later. >.<
- Mar 20, 2000
- 102,402
- 8,574
- 126
ace, those are pipeline stages you're talking about. the more a chip has, the less it has to do per cycle to process an instruction, the faster it can run. the way to get around that sort of phenomena is to increase the number of pipelines. but that has quickly diminishing returns since later instructions are often dependent on data being processed. as long as the chip guesses the outcome right, its fine, but when it misses it would have to clear the pipes and start all over. expect willy to be slower than a p6 per cycle, so it will have to have more cylces to beat one. so it will start well faster than 1GHz, since that would put it in p6 territory and show that p6s are faster. the p6 didn't have this prob when it was replacing the p5, if you recall, making the p6 intel's most powerful/cycle ia-32 chip.
We don't know for certain how well P4 will perform in IPC - yes by conventional wisdom the 20 stage pipeline is absolutely horrendous when it comes to branch mispredictions, but something called "trace cache" is supposed to fix that.
Time will tell...
Time will tell...
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 23K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter