I don't see the irony. How did I "taunt"? You don't have to agree with my reasons for not overclocking a mac pro. The reasons are "viable" though.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Why so little overclocking?
- Thread starter Loa
- Start date
- Sort by reaction score
I don't see the irony. How did I "taunt"? You don't have to agree with my reasons for not overclocking a mac pro. The reasons are "viable" though.
On the PC side of things, you can't much overclock the Xeon on Xeon motherboards either.
As far as the i7 goes, my understanding is that the i7 Bloomfield chips are are the same. The ones that test out better become the Extreme and higher clocked CPUs. What this means is that if Intel is on their game, and getting good yields, then very good quality CPUs end up become the i7 920. This is part of why they overclock so well.
As far as instability, I've never much had a problem? I hear the i7 920 can get to 4GHz, but, I've never taken it above 3.6 and usually stay in the 3.2-3.4 range. The only thing I notice is that my rendering and compression times drop. lol
I change one setting in the BIOS, and bam, new higher speeds. My last dekstop, an AMD 3800+ was also as simple to overclock, with no issues that I am aware of other than a faster computer.
As far as the i7 goes, my understanding is that the i7 Bloomfield chips are are the same. The ones that test out better become the Extreme and higher clocked CPUs. What this means is that if Intel is on their game, and getting good yields, then very good quality CPUs end up become the i7 920. This is part of why they overclock so well.
As far as instability, I've never much had a problem? I hear the i7 920 can get to 4GHz, but, I've never taken it above 3.6 and usually stay in the 3.2-3.4 range. The only thing I notice is that my rendering and compression times drop. lol
I change one setting in the BIOS, and bam, new higher speeds. My last dekstop, an AMD 3800+ was also as simple to overclock, with no issues that I am aware of other than a faster computer.
Thanks for your input.... but I'm curious as to why having dynamic logic would have a greater electromigration effect. I'm sorry, my knowledge of circuits is rudimentary at best (i.e., 200-level physics).
Also, (pardon me for being a layman) I would assume that as dies continually shrink, the power density being put through them would have to decrease in order to balance things (Black's equation) out? I understand that Black's equation isn't real life, but I assume that all other factors being the same both have to trend downward.....
As you probably know, the costs of some of the pc counterparts are equally as high if not higher than a mac pro and that never stopped an overclocker from messing about. The zdnet tool works depending on how good your ram is I've come to find out, but I've also resigned myself to the fact that if I really wanted to overclock for whatever reason, it is far easier on a pc or even a mac clone and not really worth the effort on a mac pro. Anyway as a long time pc overclocker, that isn't why I bought a MP. What most mac users don't understand is that pc overclocking is a hobby. It's fun. It can be challenging and you learn a great deal along the way. I'm not sure why some people get so worked up about the dangers and reasons we do it. No one is forcing them to even read this. Oh well......
In static CMOS the current tends, over time, to be zero. In other words, for every cycle in which, for a given wire, current flows in one direction, in another cycle it will flow in the opposite direction.
In some types of dynamic circuits this is not the case- current always flows in the same direction.
Electromigration occurs in the direction opposite current flow, as electrons essentially carry away molecules of the metal. If you have bidirectional current, electromigration is much less of a problem as the net effect cancels out- before my molecule gets very far, the current reverses and brings it back.
Yes, as metal dimensions decrease we worry much more about electromigration - but usually there aren't too many wires susceptible and we beef them up with lots of vias, increased width, etc.
Tutor - the old, partially wimpy, paradoxical contrarian.
Why overclock if you can underclock your way to the highest performance. I used to only be a stay-within-spec-VID overclocker (still am a stay-within-spec-VID overclocker only where all attempts at underclocking fails). But I have now discovered the rewards that underclocking and turbo biasing bring: 1) Keeps those voltages within spec VID. 2) Keeps system and CPU (and cores within) cooler. 3) Keeps power company at bay, lower current draw. Upon load, aren't these the relevant factors determining whether turboboosting occurs? Turbo biasing now at DDDDEE (13,13,13,13,14,14 [27 - 13 = 14; 26 - 13 = 13]), with HPET 64, C1E, C-states, P-states, T-state and all other native power management features fully enabled, running dual 5680s rated at 3.33 GHz, but running them at under 2.5 GHz. Click URL in sig, then click/filter scores, comparing scores with frequencies. All systems are completely stable - both O'ced and U'ced ones. And as for that knock against overclockers using the longevity/stability card, my overclocked 1) Atari TT030, 2) Amiga 500, 2000s, 3000 [with Video Toaster] and 4000 [with Video Toaster], and 3) Apple Power PC 7500, 8500s, 9600 are all still running fine in my studio. Getting over 17 years of regular use from these modified antiques counters the longevity/stability arguments of those who have never gotten to know their system's potential fully. And yes, I have two overclocked iMac (the original ones with those weird colors) and three Mac Pro 2,1's rated at 3.0 Ghz, overclocked to 3.6-3.7 GHz, using 800 MHz memory and a modified version of Zdnet's utility and expect to have them all, as well as all of my other post-2000 modified systems that I don't gift, running stably when I'm in my 70's - twelve years from now.
Why overclock if you can underclock your way to the highest performance. I used to only be a stay-within-spec-VID overclocker (still am a stay-within-spec-VID overclocker only where all attempts at underclocking fails). But I have now discovered the rewards that underclocking and turbo biasing bring: 1) Keeps those voltages within spec VID. 2) Keeps system and CPU (and cores within) cooler. 3) Keeps power company at bay, lower current draw. Upon load, aren't these the relevant factors determining whether turboboosting occurs? Turbo biasing now at DDDDEE (13,13,13,13,14,14 [27 - 13 = 14; 26 - 13 = 13]), with HPET 64, C1E, C-states, P-states, T-state and all other native power management features fully enabled, running dual 5680s rated at 3.33 GHz, but running them at under 2.5 GHz. Click URL in sig, then click/filter scores, comparing scores with frequencies. All systems are completely stable - both O'ced and U'ced ones. And as for that knock against overclockers using the longevity/stability card, my overclocked 1) Atari TT030, 2) Amiga 500, 2000s, 3000 [with Video Toaster] and 4000 [with Video Toaster], and 3) Apple Power PC 7500, 8500s, 9600 are all still running fine in my studio. Getting over 17 years of regular use from these modified antiques counters the longevity/stability arguments of those who have never gotten to know their system's potential fully. And yes, I have two overclocked iMac (the original ones with those weird colors) and three Mac Pro 2,1's rated at 3.0 Ghz, overclocked to 3.6-3.7 GHz, using 800 MHz memory and a modified version of Zdnet's utility and expect to have them all, as well as all of my other post-2000 modified systems that I don't gift, running stably when I'm in my 70's - twelve years from now.
Thanks for the resurrection.
Not sure what this means either. GoMac claimed in this thread that Xeons are not meant to overclock. WTF? Some of the best overclocks I have ever seen were on Xeons. You can overclock a Xeon Hex to at least 5.4GHz in right environment. I saw a a W3690 hit 6.1GHz. Intel will not be selling a default 4GHz clock for at least a year. Their turbo bins may hit it but they are holding steady at 3.9GHz for turbo even through Ivy Bridge.
Truly you have to be real timid to be scared of overclocking and ruining the chips. You may ruin the board before you ruin the chip. Overclocking has tremendous advantages. Like paying 200.00 for a core 2 duo that can stably clock beyond the 1000.00 part. I had a conroe 2.66GHz that ran for 3+ years at 3.66GHz and it was locked! Made huge difference in my game smoothness. Actually that proc is still going strong with a milder OC for my father at 3.06GHz. It is kind of like a car, just bang it into submission.
Oh, I gotta know how you can underclock and volt a Nehalem- is it Reggie?
I've heard of overclocking one by cutting a PLL clock gen (I think). Gotta tell how!
I'm intrigued by this, as I too have a 2,1 @ 3GHz and would love to hear more about getting 800MHz memory and modified Zdnet tool. Care to explain a little on how you got that set up?
Bought from Newegg the 800 MHz memory they recommend for the Mac Pro 2008. As for the software, simply download it and first try it as is because mileage varies. If you can reach their limit stably, then PM me.
Caveat: I have decided to stick with SL 10.6.7 for all of my systems that handle SL, except for one of my 2009->2010 Mac Pros and one of my MBpros. So I'd recommend that you first download the software from ZDnet before investing in the memory to see if you can run the software with whatever version of OSX you have.
----------
I've sent a private message (PM) to you.
I think it was said that with Xeon motherboards which is probably true (super micro, Tyan etc)
I overclock my Mac Pro 2008. I have been overclocking for 2 years at 3.15ghz -no problems at all, no crashes, 3day renders complete fine. This has saved me days of time & on occasion it was the only way to achieve my deadlines & It also improves the speed of my graphics card.
@ drrbc 4ghz ? -this on mac pro 1.1 tell me more.. Just got my x5482's trying for 3.55 on my 3.1 Pro & have been looking at water cooling. What blocks do you use ?
@ drrbc 4ghz ? -this on mac pro 1.1 tell me more.. Just got my x5482's trying for 3.55 on my 3.1 Pro & have been looking at water cooling. What blocks do you use ?
Well, it should be easy to figure out if ya look at my sig- but any way...
I got a 1,1 with 5350's BSEL'd to 3.3 (using Koolance CPU 340's). VID'd to 1.4125- I think. Will try the 1600 BSEL with new 800Mhz RAM (shout out to Grant at OWC). if that don't work then will try Netkas' 1,1 to 2,1 utility. If not that then I'm dropping them into a 3,1 to see what happens.
The 3,1 is to be my new main gizmo. Got some 5460's for it. See no reason why can't get 3.8 outta it.
Main thing holding me back currently is time to fabricate the waterblock's for the RAM. Havent even started the PSU yet, but that should be easier than the RAM (famous last words- ZAP!)
Bad need of expanding the GHX I'm currently using. Last on the list is to cool two 30" ACD's. I'm expecting to about double the watts into the dirt to @ 2200. In bad need to find a way to get some 10' sections of copper pipe 20' straight down on the cheap. Any PetE's or ME's in the Houston area willing to work for beer?
Later, I've got parts for a sub $2k 5,1 "FrankenMac". But not as interesting to me as the 3,1- but should be an attractive machine anyway. Hope to Geekbench 20k on it.