rraehal
08-29-2003, 03:37 PM
I have the 865 PERL Board and a Celeron. By the time the software comes out I will have P4C Hyper Threading processor. I will be getting this software to see what it can do.
Click to See Complete Forum and Search --> : Intel Readies Overclocking Software
Someone Stupid
08-29-2003, 07:34 PM
"Quiet Mode" and "Performance Mode" This sounds like software designed to shut the prescott up, not overclocking software.
causticVapor
08-30-2003, 12:18 AM
SS, I think you're right. Rather an underclocking tool - the last one to want youto get 3.2C performance from a 2.4C is intel.
"While it will be available to performance enthusiasts, the new tool is aimed at system integrators."
Sounds like a devious way to make a 3.4GHz prescott that'll run at 60% clock speed most of the time.
"While it will be available to performance enthusiasts, the new tool is aimed at system integrators."
Sounds like a devious way to make a 3.4GHz prescott that'll run at 60% clock speed most of the time.
Someone Stupid
08-30-2003, 09:49 PM
cV: Are you the one who had told me (and I think you were) that the prescott was going to need 100 watts, then mentioned a die shrink to .09 microns wouldn't work as it would still leave you with more than the 30% loss of heat you would need? If so, I read an article on Overclockers.com a decent while back, and one somewhere else (sorry, link escapes me) that the 100 watt rating was for a 0.09 micron processor - not the 4 ghz rated one, but a step below it (it was based on projections, but they reached the same conclusion nearly). The 4 ghz would need btw 100 to 110 watts going by predictions, the 3.2 would be around 75 to 80 IIRC right - it's been awhile. Granted it's 64 bit internal, but with 32 bit external bus, that wipes that off unless they get hyperthreading to take advantage of that somehow - but you still have the bottleneck, and you get 12 new instruction sets, just you have that heat to deal with. Are those 12 sets going to be enough to justify a prescott purchase to most overclockers or just high end enthuiest? Not the overclockers, no headroom, the enthusiast, I doubt as well - noise. So if AMD can keep their puppy cool and running well, they have a winner again.
Looks like Mosfet's are going to need to be actively cooled well and given more space now. If you go to more mosfets, your taking up more board space and increasing the cost as you need more still pumping high voltages, if you go to fewer or the standard 4 set, your looking at having to up the quality and specs of them, thus also increasing the power they draw. If they want to continue with raising clock speeds (AMD and Intel), eventually dual sided heatsinks (active) will be needed on boards and maybe even some stock watercooling. The Prescott may make it under with it's high requirements with slight changes to current setups, but the next generation afterwards, even with another die shrink - if they keep pushing for faster, cooling is going to become that much more important - hell - might even change the way PCB's are made if your looking at sending a chip through it due to heat.
If they keep going for faster - which we are hitting a limit as to what we really can do with those clockcycles on a home rig, even for gaming - it can't get too much better graphically, I mean look at quake 3 to doom 3 engine wise on a graphics and other programming levels. Now tell me you don't think the engine after that which carmack will release, IF it's made for a couple years from now will be of, it not nearly of virtually cinematic quality with even better physics? It's nearing it now with the help of vid cards and processing power right now, as that is carmack's style, but if comps don't get some real increases in speed, carmack's next engine will be a computer killer more so than Doom3 was/is touted to be. That is about the only thing that PCs need the speed for these days - workstations and servers it's a bit different - even in servers, just adding another rack of processors may be easier than overcoming oncoming heat problems - workstations you can only go quad 4's with hyperthreading, and it will take a big board with some creative cooling to keep it quiet if Intel can't get heat under control - AMD will run into the heat issue again, it's not like they really are behind intel, it's just at their PR rating they are ahead in keeping chips cool - in benches to test them, your looking at AMD's still being hot chips, just as intel's are now hot chips. Can't undermine simple physics. More current in a smaller space means more heat and less area to dissipate it.
I wouldn't mind the CPU running "intelligently" and throttling up when needed, but I want it seemless as most apps don't need even a gig of processor power to run and open within a reasonable time (barring games and workstation apps). Though when I need the power, I just want it to go, no waiting and whatnot and certainly not on some pissy program, it's either chip driven instructions or a logic controller on the board. I also don't want to hear my computer turn into a blowdryer like with how they did it on the 5600fx or whatever it was that was on the shelves for less time than it was marketted practically.
So it looks like it's time to address heat issues, not speed issues very soon. For intel that reality is now, for AMD, a little later, not much more unless they have some serious tricks Intel doesn't know (not likely).
I can see Intel chips being marketted slightly below their actual rated speed - then having a built in locked speed (like the multiplier now) on next gen chips if it takes them to use the program and thus the overclocking your doing is actually just clocking it to what it was supposed to run but they call it overclocking :rolleyes: - though you pay for that in dbs from a fan it sounds like with Intel's "overclocking" sales pitch to the gullible.
On both the Barton and Northwood core, when you see it pass 2.6 (northwood) to 1.83 (barton)you really start to see wattages ramp up a lot more than they did on slower speeds for each megahert increase. I just think we are pushing technology at stock on the upper end chips more than what science really allows with current technology. We can do it safetly, just the wattage jump shows we are starting to get diminishing returns and badly at that... time for RnD to mean RnD.
I like both AMD and Intel - depends on what I have to build the system for, and the budget - but I see a crisis looming in Intel's camp that either AMD doesn't have, or is keeping under wraps.
Looks like Mosfet's are going to need to be actively cooled well and given more space now. If you go to more mosfets, your taking up more board space and increasing the cost as you need more still pumping high voltages, if you go to fewer or the standard 4 set, your looking at having to up the quality and specs of them, thus also increasing the power they draw. If they want to continue with raising clock speeds (AMD and Intel), eventually dual sided heatsinks (active) will be needed on boards and maybe even some stock watercooling. The Prescott may make it under with it's high requirements with slight changes to current setups, but the next generation afterwards, even with another die shrink - if they keep pushing for faster, cooling is going to become that much more important - hell - might even change the way PCB's are made if your looking at sending a chip through it due to heat.
If they keep going for faster - which we are hitting a limit as to what we really can do with those clockcycles on a home rig, even for gaming - it can't get too much better graphically, I mean look at quake 3 to doom 3 engine wise on a graphics and other programming levels. Now tell me you don't think the engine after that which carmack will release, IF it's made for a couple years from now will be of, it not nearly of virtually cinematic quality with even better physics? It's nearing it now with the help of vid cards and processing power right now, as that is carmack's style, but if comps don't get some real increases in speed, carmack's next engine will be a computer killer more so than Doom3 was/is touted to be. That is about the only thing that PCs need the speed for these days - workstations and servers it's a bit different - even in servers, just adding another rack of processors may be easier than overcoming oncoming heat problems - workstations you can only go quad 4's with hyperthreading, and it will take a big board with some creative cooling to keep it quiet if Intel can't get heat under control - AMD will run into the heat issue again, it's not like they really are behind intel, it's just at their PR rating they are ahead in keeping chips cool - in benches to test them, your looking at AMD's still being hot chips, just as intel's are now hot chips. Can't undermine simple physics. More current in a smaller space means more heat and less area to dissipate it.
I wouldn't mind the CPU running "intelligently" and throttling up when needed, but I want it seemless as most apps don't need even a gig of processor power to run and open within a reasonable time (barring games and workstation apps). Though when I need the power, I just want it to go, no waiting and whatnot and certainly not on some pissy program, it's either chip driven instructions or a logic controller on the board. I also don't want to hear my computer turn into a blowdryer like with how they did it on the 5600fx or whatever it was that was on the shelves for less time than it was marketted practically.
So it looks like it's time to address heat issues, not speed issues very soon. For intel that reality is now, for AMD, a little later, not much more unless they have some serious tricks Intel doesn't know (not likely).
I can see Intel chips being marketted slightly below their actual rated speed - then having a built in locked speed (like the multiplier now) on next gen chips if it takes them to use the program and thus the overclocking your doing is actually just clocking it to what it was supposed to run but they call it overclocking :rolleyes: - though you pay for that in dbs from a fan it sounds like with Intel's "overclocking" sales pitch to the gullible.
On both the Barton and Northwood core, when you see it pass 2.6 (northwood) to 1.83 (barton)you really start to see wattages ramp up a lot more than they did on slower speeds for each megahert increase. I just think we are pushing technology at stock on the upper end chips more than what science really allows with current technology. We can do it safetly, just the wattage jump shows we are starting to get diminishing returns and badly at that... time for RnD to mean RnD.
I like both AMD and Intel - depends on what I have to build the system for, and the budget - but I see a crisis looming in Intel's camp that either AMD doesn't have, or is keeping under wraps.
causticVapor
08-31-2003, 10:53 AM
I looked closely at the intel overclocking program, and based on the screenshot, it seems that "400MHz DDR/800MHz FSB" is the highest setting. The two speedometer things indicate the psuedo-dual-processor usage, and seemingly a 217 something is at the bottom. Now, I've done some simple math and it seems that in order to get 3000MHz at 217 FSB, they would need to use a 2.7C. But there is no 2.7C, and the 2.8C runs at 3038MHz on 217FSB. :confused:
I agree with you on the heat issue; many sources are already bringing to light (overclockers.com, aceshardware.com) that we're getting less and less of a % speed increase for disproportionately more power consumption, especially at the tail end of technologies.
http://img.villagephotos.com/p/2003-8/339670/wattages.gif
Ed made a statement in one of his articles and I've highlighted the general sentiment here in this graph of mine - in that prescott bucks the trend of less power consumption at the same MHz.
Voltage regulator cooling is nice, but eventually six-phase power will be required (or larger 4-phase regulators, taking up more space on the board.)
I agree with you on the heat issue; many sources are already bringing to light (overclockers.com, aceshardware.com) that we're getting less and less of a % speed increase for disproportionately more power consumption, especially at the tail end of technologies.
http://img.villagephotos.com/p/2003-8/339670/wattages.gif
Ed made a statement in one of his articles and I've highlighted the general sentiment here in this graph of mine - in that prescott bucks the trend of less power consumption at the same MHz.
Voltage regulator cooling is nice, but eventually six-phase power will be required (or larger 4-phase regulators, taking up more space on the board.)
causticVapor
08-31-2003, 02:30 PM
Higher wattage numbers tend to mean a hotter-running CPU, but that is not always the case. The number of watts a CPU releases per square millimeter determines how hard a CPU is to cool.
http://img.villagephotos.com/p/2003-8/339670/watts_per_sq_mm.gif
This shows that a Prescott running at 4GHz on its current stepping would be roughly twice as difficult to cool as a Thunderbird 1400. Notice how cool the Willamette 2GHz is.
http://img.villagephotos.com/p/2003-8/339670/watts_per_sq_mm.gif
This shows that a Prescott running at 4GHz on its current stepping would be roughly twice as difficult to cool as a Thunderbird 1400. Notice how cool the Willamette 2GHz is.
Someone Stupid
09-01-2003, 03:03 AM
And we all know how cool those thunderbird chips ran. :) Now shrink the surface area and let the smoking begin.
cV: The 2.7 may actually be the new 2.8 prescotts default speed, along with a 400 bus if things are really bad. If not it could be the same but on an 800 FSB. Or it could be an entirely new chip or some BS PR shot with no relevance whatsoever. Wouldn't be the first time, it won't be the last if it is.
cV: The 2.7 may actually be the new 2.8 prescotts default speed, along with a 400 bus if things are really bad. If not it could be the same but on an 800 FSB. Or it could be an entirely new chip or some BS PR shot with no relevance whatsoever. Wouldn't be the first time, it won't be the last if it is.
SysOpt.com
Copyright Internet.com Inc. All Rights Reserved.
Copyright Internet.com Inc. All Rights Reserved.