BonzaiDuck
Lifer
- Jun 30, 2004
- 15,882
- 1,550
- 126
FOOTNOTE:
"Pedro" is using a P35 motherboard, only recently released.
Three months ago, I visited Alfredo Comparetti's web-site for SpeedFan. The version just released at that time was the last version he planned to release. Similarly, "Serj," who designed SnM, has stopped revising his software. Finally, Everest Ultimate Edition must also strive to keep up with new motherboards.
But this was a problem for SpeedFan, as users of nVidia chipsets well know. I also find it odd that Pedro doesn't post results with CoreTemp, but then CoreTemp doesn't run under VISTA.
So again, the only set of temperatures that are reliable in the Xtreme thread are the BIOS temperatures. These would only be IDLE temperatures biassed upward by some few degrees C, and the CPU (TCase) value shown there is 52C. At full load, even the G0's TJunction temperatures would be higher than Tcase by 10 to 15C -- or some similar margin. But the TCase value at idle is already coming within 11 to 13C degrees of the throttling spec posted by Intel.
I advise people to be careful what they do with their hardware and their hard-earned cash. NO AIR-COOLER, NOR ANY WATER-COOLER, NOR EVEN PHASE-CHANGE would mitigate the degrading effect of setting the VCORE to 1.6+V when the spec limit is clearly 1.35V or less. You would be fooling yourself with temperatures, and in this case, with temperatures that are totally -- and logically -- unreliable.
And once again -- you do not need to over-clock a system or set up a benchtest configuration at other than stock settings, to COMPARISON TEST a pair of coolers. The Coolers' effectiveness will prove at whatever processor thermal wattage is set in the comparison. The Anandtech comparison reviews, including that of May 4, simply use a fixed set of over-clock settings to prove what would be logically evident from the stock settings, except that those reviews actually show the limits that can be reached for a particular test-bed configuration.
But to take a single cooler -- like the Sunbeam Tuniq -- then boost the VCORE to 1.6+V for the recently-released G0, and say that this "proves" something about the Tuniq -- is pure baloney. It only proves that you can boost the VCORE of a G0-stepping Q6600 to 1.6V and get it to show stable results under ORTHOS for some several minutes (maybe -- even -- a whole hour!) If a comparison review between a Tuniq and another cooler using ANY processor or common test-bed configuration, and ANY over-clock setting common between the two cooler tests, shows that the Tuniq slips past the finish line as a close second, it will slip past the finish line at a close second with a VCORE of 1.6+V, using a newer motherboard, dated and unreliable temperature-measurement software, and bad judgment about what the processor's manufacturer actually intended or what the technology is capable of sustaining in terms of voltages.
To conclude -- some people need to go back to school and study logic. There is something called the Transitivity Rule: "If B is greater than A, and C is greater than B, then C is greater than A."
And that's why a review that isn't a comparison review has limited usefulness, even if you can trust the person doing the tests to be careful, objective and scientific in controlling and executing those tests. No offense intended, but "Who is Pedro?"
"Pedro" is using a P35 motherboard, only recently released.
Three months ago, I visited Alfredo Comparetti's web-site for SpeedFan. The version just released at that time was the last version he planned to release. Similarly, "Serj," who designed SnM, has stopped revising his software. Finally, Everest Ultimate Edition must also strive to keep up with new motherboards.
But this was a problem for SpeedFan, as users of nVidia chipsets well know. I also find it odd that Pedro doesn't post results with CoreTemp, but then CoreTemp doesn't run under VISTA.
So again, the only set of temperatures that are reliable in the Xtreme thread are the BIOS temperatures. These would only be IDLE temperatures biassed upward by some few degrees C, and the CPU (TCase) value shown there is 52C. At full load, even the G0's TJunction temperatures would be higher than Tcase by 10 to 15C -- or some similar margin. But the TCase value at idle is already coming within 11 to 13C degrees of the throttling spec posted by Intel.
I advise people to be careful what they do with their hardware and their hard-earned cash. NO AIR-COOLER, NOR ANY WATER-COOLER, NOR EVEN PHASE-CHANGE would mitigate the degrading effect of setting the VCORE to 1.6+V when the spec limit is clearly 1.35V or less. You would be fooling yourself with temperatures, and in this case, with temperatures that are totally -- and logically -- unreliable.
And once again -- you do not need to over-clock a system or set up a benchtest configuration at other than stock settings, to COMPARISON TEST a pair of coolers. The Coolers' effectiveness will prove at whatever processor thermal wattage is set in the comparison. The Anandtech comparison reviews, including that of May 4, simply use a fixed set of over-clock settings to prove what would be logically evident from the stock settings, except that those reviews actually show the limits that can be reached for a particular test-bed configuration.
But to take a single cooler -- like the Sunbeam Tuniq -- then boost the VCORE to 1.6+V for the recently-released G0, and say that this "proves" something about the Tuniq -- is pure baloney. It only proves that you can boost the VCORE of a G0-stepping Q6600 to 1.6V and get it to show stable results under ORTHOS for some several minutes (maybe -- even -- a whole hour!) If a comparison review between a Tuniq and another cooler using ANY processor or common test-bed configuration, and ANY over-clock setting common between the two cooler tests, shows that the Tuniq slips past the finish line as a close second, it will slip past the finish line at a close second with a VCORE of 1.6+V, using a newer motherboard, dated and unreliable temperature-measurement software, and bad judgment about what the processor's manufacturer actually intended or what the technology is capable of sustaining in terms of voltages.
To conclude -- some people need to go back to school and study logic. There is something called the Transitivity Rule: "If B is greater than A, and C is greater than B, then C is greater than A."
And that's why a review that isn't a comparison review has limited usefulness, even if you can trust the person doing the tests to be careful, objective and scientific in controlling and executing those tests. No offense intended, but "Who is Pedro?"