I agree that manufacturers have stated it, but I have yet to see Intel state anything about it.....My point is I don't think they promised anything and many were already speculating new power requirements may have ruled out many boards anyways...
I agree that manufacturers have stated it, but I have yet to see Intel state anything about it.....My point is I don't think they promised anything and many were already speculating new power requirements may have ruled out many boards anyways...
Even if Intel hasn't officially announced compatibility (I thought they had, but I could be mistaken) it has to be assumed that the 875 canterwood would at least have to support the Prescott, otherwise what good is it to launch the processor without a chipset to support it.
edit: motherboards on the other hand could be questionable, but I think its safe to say chipset support is there.
Actually I agree...i just saw Intel's comment in another thread but the mobo support is the big deal now....Unless the manufacturer is going to update bios correctly identifying the prescott and all the features may be questionable....Vcore may be too high as well if it doesn't recognize it and sets it to northwood defualt of 1.525v....This thing doesn't need anymore reasons to be hotter!!!!
There is no way Intel has lost with the Prescott. I had hoped it would go big right from the get go, but it seems now that Prescott = 755 and most everything Prescott now is all about branding/positioning Prescott. For example, post-release I can't even find anywhere to get one. But who knows, maybe there's some good 478 Prescott potential to squeeze out. We'll just have to wait and see.
Right now is a tough time because A64 has momentum but needs (IMO) some fresh chipset/mobo blood and Prescott's potential will become clearer on its home mobo turf. And then there is all the PCI-ex, DDRII, etc. stuff coming. I am going to upgrade (build a second home system actually --> current system to wife), and I hope to squeeze in best performance at my price point without basically buying everything new. A64 looks good right now, but I definately won't rule out Prescott (or maybe even NW) even with all the recent "bad" press. Life's too short, my money's too limited, and AMD/Intel performance is too similar to play the brand loyalty game. Ultimately I hope AMD and Intel stay pretty much even round after round. If in some bizarre and very much unlikely twist of fate, the Prescott does lemon out, we might be the losers (in the short term). AMD can use the cash inflow (they aren't Intel!) and wouldn't have incentive to sell at lower prices, I suppose. In the longer term Intel is a formidible cpu force, and I would have a hard time betting against them in any way, even if I do have a little bit of a go underdog sentiment for AMD.
Right now is a tough time because A64 has momentum but needs (IMO) some fresh chipset/mobo blood and Prescott's potential will become clearer on its home mobo turf. And then there is all the PCI-ex, DDRII, etc. stuff coming. I am going to upgrade (build a second home system actually --> current system to wife), and I hope to squeeze in best performance at my price point without basically buying everything new. A64 looks good right now, but I definately won't rule out Prescott (or maybe even NW) even with all the recent "bad" press. Life's too short, my money's too limited, and AMD/Intel performance is too similar to play the brand loyalty game. Ultimately I hope AMD and Intel stay pretty much even round after round. If in some bizarre and very much unlikely twist of fate, the Prescott does lemon out, we might be the losers (in the short term). AMD can use the cash inflow (they aren't Intel!) and wouldn't have incentive to sell at lower prices, I suppose. In the longer term Intel is a formidible cpu force, and I would have a hard time betting against them in any way, even if I do have a little bit of a go underdog sentiment for AMD.
Even if you does overclock well with slight voltage bump, just imagine the obscene amount of heat that flamethrower is putting out!
SO far from what I can tell ther is no need to increase voltage to take a 2.8 prescott to the 3.5 area and 4 is an easy bet with about half a volt more. And ocer's know that half a volt will usally not affect temps more than 8C even with stock cooling.
8C from Xbitlabs' estimated 61C full load temps would come out to be a disgusting ~70C, that's unacceptable.
If you paid attention you would have noticed that was with a retail hsf....No one should do vcore boosting ocing with a retail hsf anyways....A person who invest in an aftermarket hsf like 32 dollars for a zalman 7000a al-cu could and should have had those temps lower by a significant amount under load...Do you have any idea how piss poor a retail hsf is??? The fan rpm is real low and the fan is only a 70mm fan...A zalman with 92mm fan put into normal mode will double to triple the cfm also while having a superior base for heat transfer.
I have seen a few prescott threads now with 3.7ghz out of a 3.2ghz with default to only .05v boost on air....
I bet on an asus mobo and a zalman or slk thermalright hsf you may see high 50's to just into 60's...That would be acceptable...
For the true aggressive ocers they don't have any qualms about a 60-100cfm tornado fan they have plenty of room to take the chip to 4ghz if it only needs .1v as I have also seen....
True aggressive OC'ers? Those guys have water and phase change.
Anyways, I was just making a parallel to what was already out there, yes I knew they were using the stock cooler. But even with an improved cooler, there will be a change between non-OC and OC'ed temps. So you go from 61C (stock cool/stock speed) to 58C (zalman/voltage bump) under load, that's still rather hot. I don't think I'm the only person here who thinks over 50C is too hot.
Yeah I see you don't know much about ocing an retail hsf...The 61c was stock cooling at stock voltage...I bet a zalman or thermalright will be low 50's at load...58c LOL!!!!!!!!!....i am saying this will buy them the ability to achieve an oc...as long as they don't raise voltage the gain of 500mhz at default voltage will not take them likely over 60c if in fact they were able to run low 50's at default.
On my crappy abit temp reporting board even at 2.4ghz stock I will load into the low 50's...oc'd to 3.4ghz with default the highest vcore I have seen is 60c...10c for 1ghz!!!! for 100mhz more and .05v I add another 4c max I have seen.....This is ofcourse running dual instances of prime95....IN most times I don't run like this or run an app 100% but deosn't generate this much heat so I usually see about 2c less at load....
there is room for some air cooling ocing for those with aftermarket coolers only....even more room for those with non sensitive ears and 60-100cfm fans....even more for water coolers...and the highest limits we have likely seen with phase changers....
On my crappy abit temp reporting board even at 2.4ghz stock I will load into the low 50's...oc'd to 3.4ghz with default the highest vcore I have seen is 60c...10c for 1ghz!!!! for 100mhz more and .05v I add another 4c max I have seen.....This is ofcourse running dual instances of prime95....IN most times I don't run like this or run an app 100% but deosn't generate this much heat so I usually see about 2c less at load....
there is room for some air cooling ocing for those with aftermarket coolers only....even more room for those with non sensitive ears and 60-100cfm fans....even more for water coolers...and the highest limits we have likely seen with phase changers....
I get mid 40's right now under load (2.4C @ 3.3GHz, default volts, Zalman 7000AlCu, warm SoCal weather), and with Xbit's delta of 13C between Northwood and Prescott, I estimated 58C from my mid 40's. Just a guesstimation as is your guesstimation of low 50's. I ain't even worried about it though, whatever.
I've heard through the grapevine that the current stepping of the prescott is only a 7 layer chip, and at later steppings its supposed to go up to an 11 layer chip. Which would dramatically alter chip the chips internal layout and can dramatically effect temps. Anyone know of any links or validity to this?
The bottom line is that heat problems WILL be worked out - either by Intel or by overclockers. Howwever, I don't really care about the heat as long as I can overclock enough to make the chip faster than a Northwood (3.6GHz+) and have a stable system.
I tried an AMD chip which didn't work out for me & so I'm not going back. Intel is going to be my solution and it's capabilities fit my needs (multi-tasking, light gaming, more video/audio usage) better. I just nee to figure out how long I want to wait to upgrade to a HT chip & whether that chip will be a Northwood or Prescott (I hope because of PCI-express & DDRII will be on the chipset that "officially" supports Prescott).
I tried an AMD chip which didn't work out for me & so I'm not going back. Intel is going to be my solution and it's capabilities fit my needs (multi-tasking, light gaming, more video/audio usage) better. I just nee to figure out how long I want to wait to upgrade to a HT chip & whether that chip will be a Northwood or Prescott (I hope because of PCI-express & DDRII will be on the chipset that "officially" supports Prescott).
MadRat
Lifer
- Oct 14, 1999
- 11,978
- 295
- 126
The bottom line is that INTEL is making a server chip marketable to the masses. Prescott is the basis for the next Xeon line. The difference bewteen P4E and Xeon-E will be 4MB of full-speed, on-die Level 3 cache memory. The core is also showing signs under a microscope that it likely has Thread-Level Parrallelism (TLP) and probably has two execution units and a separate D-cache for each unit. No need for raw clock speed when you have so much TLP potential. The layout would make the Xeon-E look like it has two separate processors. Add in the SMT they use, called hyper Threading, and it makes the chip to appear to have four cores. Not too shabby.
So where did Intel go wrong?
So where did Intel go wrong?
Looks like Prescott is useful if you use HT...
Prescott Hyper-Threading Examined Conclusion
"From what we?ve seen, the Prescott occasionally gains up to an additional 4% with Hyper-Threading than Northwood did. "
"For those of you who doubt Hyper-Threading, use an HT enabled Pentium 4 and Athlon 64, Athlon XP, or even non HT Pentium 4 for an extended period of time, it does make a difference."
Prescott Hyper-Threading Examined Conclusion
"From what we?ve seen, the Prescott occasionally gains up to an additional 4% with Hyper-Threading than Northwood did. "
"For those of you who doubt Hyper-Threading, use an HT enabled Pentium 4 and Athlon 64, Athlon XP, or even non HT Pentium 4 for an extended period of time, it does make a difference."
Some French site
Use babelfish to translate...
When overclocked to 4GHz, the chip was pretty much a close second to the P4EE chip overclocked to 3.7GHz
More interesting is what they found:
"The invisible part is now what is really Prescott. It appears clear that the study of the die has, the prescott is almost a new architecture and not a simple evolution like was Northwood. The presence proven and proven of a Trace Hiding place larger than than INTEL wants to say to us well, and especially, of what resembles two cores of execution leaves thoughtful. One thus sees very clearly two blocks of 16 KB memory hiding place L1 whereas the specifications official make well on state of only 16 KB of memory. If the activation of these technologies were probably not shoed before Tejas, it could be activated well in Xeon "Prescott" which will be available soon. At all events, it is clear that Prescott has much thing has to reveal..."
Use babelfish to translate...
When overclocked to 4GHz, the chip was pretty much a close second to the P4EE chip overclocked to 3.7GHz
More interesting is what they found:
"The invisible part is now what is really Prescott. It appears clear that the study of the die has, the prescott is almost a new architecture and not a simple evolution like was Northwood. The presence proven and proven of a Trace Hiding place larger than than INTEL wants to say to us well, and especially, of what resembles two cores of execution leaves thoughtful. One thus sees very clearly two blocks of 16 KB memory hiding place L1 whereas the specifications official make well on state of only 16 KB of memory. If the activation of these technologies were probably not shoed before Tejas, it could be activated well in Xeon "Prescott" which will be available soon. At all events, it is clear that Prescott has much thing has to reveal..."
I'm pretty sure the Max3 will support it-I just saw a new BIOS (21) for my Max2
Since the Max3 is the "new, improved" version I figure a new BIOS should be released shortly.
Mike G
LOL funny conclusion is that, I'd say. Let's see their results (with his own commentary in each one):
Sisoft Sandra:
In Sandra?s CPU test, the 3E receives a 37% improvement in FPU and 5% in ALU performance when HT is enabled.
The 3C receives a 40% FPU boost and a 22% ALU boost when HT is enabled.
When Sandra Multimedia is at question, the 3E receives a 32% FPU boost and 20% ALU boost.
In comparison with the 3C?s 33% FPU and 19% ALU boost, they both receive about the same amount of increase with Hyper-Threading.
Memory bandwidth is totally unaffected by Hyper-Threading and is a chipset issue for the most part.
Sysmark 2004:
Overall in Sysmark 2004, the 3E and 3C obtain a similar 11% score increase when Hyper-Threading is enabled.
MPEG2 Encoding:
When it comes to MPEG2 encoding, the 3E, while being faster than the Northwood, only receives a 20% gain with Hyper-Threading while the similarly clocked Northwood obtains a nicer 22% gain.
Lame MP3 Encoding:
Using the LAME MP3 Encoder, Hyper-Threading does nothing for performance for Prescott and Northwood.
(Really, looking at their Lame results, Northwood gains a 2%; Prescott loses a 2%)
Cinebench 2003:
When 3D rendering is at question, the 3E obtains a 17% gain when processing multiple threads. Here the 3C gains a smaller 16% gain when multiple threads are assigned.
PCmark2004:
PC Mark evaluates a system?s total performance using various tests. With the Prescott, the addition of Hyper-Threading increases the score by 12% while Northwood gains a smaller 8% with HT.
WOW finally Prescott gains up to an additional 4% with Hyper-Threading than Northwood. Pitifully it's only a synthetic bench and not real app :-( And with the other synthetic Sandra, Prescott losses HT perfomance.
In real apps they tests (in games there isn't HT impact), equal with apps in Sysmark suite; in MPEG encoding Prescott losses a 2% HT perfomance respect to Northwood; in Lame MP3 encode Prescott losses a 4% HT perfomance respect to Northwood; and in Cinebench rendering Prescott and Northwood really equal (Prescott gains exactly a 16,57% with HT and Northwood gains exactly a 16,19% with HT).
So really funny conclusion by accelenation reviewer lol
Still no final word on increases in HT, but this review also indicates an increase. I think we'll see more of the facts come out in the next couple of weeks, but it looks promising.
Hyper-Threading Improved?
"In Lightwave 3D 7.5, the Northwood cores consistenly produced a 10% speed gain with Hyper-Threading. However, when we ran the same benchmark with a Prescott core, the speed gains were immediately boosted to about 16%."
"Likewise in Cinebench 2003, we saw Hyper-Threading performance gains improving only very slightly from about 18% to 19-20%. While it may not seem like much, the results here are certainly quite encouraging since these benchmarks are based on relatively old applications which have not even been optimized for HT technology. "
Hyper-Threading Improved?
"In Lightwave 3D 7.5, the Northwood cores consistenly produced a 10% speed gain with Hyper-Threading. However, when we ran the same benchmark with a Prescott core, the speed gains were immediately boosted to about 16%."
"Likewise in Cinebench 2003, we saw Hyper-Threading performance gains improving only very slightly from about 18% to 19-20%. While it may not seem like much, the results here are certainly quite encouraging since these benchmarks are based on relatively old applications which have not even been optimized for HT technology. "
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 Discussion Threads
- Started by Tigerick
- Replies: 22K
-
-
Facebook
Twitter