Processor upgrade 1.1 macpro. is it worth it?

Wow such a fast reply was the last thing I expected to see, but it's a very welcome thing indeed, thanks!

Just for comparisons sake, I'm not running any RAID anythings. My graphics card is an original (as far as I can tell) Radeon X1900. Only change is that I upgraded the stock cooler years ago. I'm also running dual monitors, DELL U2711 and DELL U2410.

I had been considering upgrading video card and SSD as a next step but given the issues I'm having I'm reluctant to throw more money at it until I know the issue can be resolved. I will be following this thread and your site very closely, very interested in what you find and hope it's all great news for you!

Scoopz are you running SMC Fan Control or similar? If you are can you post your idle CPU temps and fan rpm etc when you get a chance? Thanks for your help/replies and all the best for your 5770 upgrade!

On Mac Proi 1.1 upgraded with X5355 processors, smcFanControl advises idle temperature is 22ºC and fan speed is 499rpm. It is winter down under and outside air temperature is currently 10º. In summer have seen temperature go to 34º and have never had to incrase fan speed.

Thanks for the info harryb. I'm also in Aus but am running much higher fan rpm to get the same 22ºC (presume this is your ambient temp?). Northbridge is at 45ºC, CPU A Heatsink 33ºC, CPU B Heatsink 29ºC. To attain these temps I need to run CPU Fan at 1200rpm, rear fan 1300rpm and HDD at 800rpm, so a great deal more than your stock 499 setting!

Triying a few things and will report back with any info if I have any. Not to steal his thunder, but for those not aware of scoopz other thread, he has succesfully installed the 5770 but still experiencing kernel panics in 10.6.8 and Lion. Surely there's other 1,1 upgraders out there having similar issues?

Currently Ambient is 13º, CPU A is 23º, CPU B is also 23º, Expansion slot 26º, HD #1 15º and HD #2 is 17º and all fans at 499 rpm including Power Supply, Exhaust, CPU and HD/Expansion slot. Regarding the kernel panics, have read this happens with faster CPU's than the 5100/5300 series.

Thinking further I do clean the fan units out regularly using a 1" paint brush and a vacuum cleaner with a rubber sticky beak to remove dust, and the Mac Pro is covered by a dust cover when not in use. It is absolutely amazing the amount of dust that gets in allthepoo, even in an air conditioned carpetted home.

Twice now I have replaced the CPU's in muy Mac Pro 1.1 and each time have removed the main fan, cleaned it out well, including wiping down the fan blades, prising up the centre seal and lubricating the fan with a couple of drops ONLY of light machine oil, reaseating the seal firmly. I do not hear any internal fan noises, only the ATI Radeon HD4870 1GB graphics card which makes a little noise on startup until the chime cuts in, and then nothing from it either.

Appreciate the additional temp info thanks. I completed my processor upgrade a few weeks ago and was a little shocked at what I discovered in terms of dust. I've built/upgraded machines in the past but never seen dust caked so thick like this was! Everything inside is now totally clean but the suggestion of a drop of machine oil is a great one.

In relation to temperatures, kernel panics etc, the main thing that frustrates me about it all is that I'm sure the X5365 was one of the original upgrade options from Apple for the 1,1 so KPs aside, given my machine is clean and ambient temps in the house are low-normal (around 20ºC), why would I be getting such high temperatures, especially when I'm running 1200rpm to get them?!

At 2pm Ambient temnperature is 21º, CPU's both 35º, Epansion 36º, both HDD bays 26º and fan speeds are Power Supply 599rpm, Exhaust 599rpm, CPU and Expansion Slots both 499rpm.

When you click on smcFanControl icon in the Menu Bar, is the 'Default' box ticked? Are you running the latest version, 2.3? This in Preferences and shows fan speed minimum settings. If your settings are the same as default, maybe slowing the fans down will have no adverse effect.

By going into FAQ, you will see it is safe to lower the settings to the default minimum as the software will increase fan speed if necessary.

I upgraded my Mac Pro 1.1 to 2 x X5355 (2.66GHz, Quad) and saw my Geekbench numbers more than double to just over 10,000 - not bad for an sub $300 upgrade. However, since upgrading to Lion I have had a couple of kernel panics and a number of minor lockups. The last full blown panic listed Little Snitch as a potential problem. I have since updated that with the newest version which 'improves Lion compatibility'. Prior to Lion, my system was bullet proof, but that doesn't mean that Lion is the cause of the problems per se. It could be that third party software is no longer compatible with the upgraded OS and is causing issues. I'll stick with Lion until after 7.1 comes out and then decide whether to go back to 10.6.8. In the meantime, it might be worth checking all your software to see if Lion compatibility updates have been issued. There are a few things to dislike about Lion, not least of which is that the scroll bars in Finder/Application windows are too thin and periodically disappear for no good reason. They don't call it the bleeding edge for nothing. Also, as readers have reported, the processor info in 'About this Mac' now reads '2.66GHz Unknown'. The current version of the utilitity 'aboutthismac.pkg' does not fix it, and I couldn't really follow the thread on kext injector. It's not a drama - I know what the processors are. Regards, BB

It is worth putting Lion on its own hard drive instead.

I have bookmarked a dozen or more Lion FAQs, compatibility tables, checker, tips.

A clean install is safer, still lets you use Setup or MIgration Assistant, and determine if a plain vanilla install works.

I'd say some elements could take 6-9 months, even 10.6 it took a year at times. Fix one, add bug, wait for graphic driver, wait for next patch or release, and even a driver can take months to test.

Get rid of Little Snitch and see how you go.

The fix is simple. Download the CPUInjector.file, open and drag CPU.Injector to Desktop. Download kext.utility, unzip and a box like icon appears on the desktop. Drag the CPU.Injector file on top and away it runs.

Runing latest SMCfancontrol 2.3. Lowest settings I can run to maintain a reasonable idle temp is 1000rpm for cpu fan, 1000rpm for exhaust, 600rpm for both power and HDD. This gives me idle temps of 33º for CPU, 48º for Northbridge and 40º-50º for various memory modules. So essentially I'm getting similar temps to your last post Harryb, but have to run much higher rpm to achieve that. Running default fan speeds results in pretty rapid rise in temperatures and inevitable auto-reset. When I push the machine hard, it does make fans spin up and keeps temps stable and doesn't necessarily reset, which to me seems a bit of a contradiction. If it's temperatures causing my resets perhaps it's because when required additional cooling isn't kicking in quick enough?

Additionally I'm now not even getting KP reports or any form of warning before restart, machine just stops responding for 1-2 secs then restarts. Once back into OS, no reports, just as if it had been started fresh. Overall I seem to be having less resarts that a few days ago (perhaps down to one a day).

Having a very hard and frustrating time figuring this out!

I'm feeling lucky... completed my upgrade early this week to x5355 2.66 GHz SLAEGs (200USD for the pair, used from ebay--insane bargain) and haven't had any problems at all. No KPs, no other problems. I run little snitch 2.4.1 (2051) without incident. OS is 10.7.2 11C26, SMC version is 1.7f10 (the updater mentioned earlier in this thread said I didn't need it). I did the CPU injector routine and the procs are reported as "quad core xeons" without the 5355 part.

The hardware monitor reported temps are fairly warm but to me not particularly alarming, but I haven't had any KPs either. At factory default fan speeds, ambient around 22C the 8 procs like to idle in the high 40Cs. Sustained full load the system kicks up the fan speeds when the cores get into the 80C range and they stay pretty stable--fan speeds never get above about 750 RPM though. I put in an SMC profile that ran CPU at 800 and rear fans at 900 then had it run overnight at full load reencoding the 20k songs in my music library. I've done this this 2 times with shutdown and cool off in between and the temps at full load were in the high 70Cs. The computer didn't need to raise the fan speeds at all. I suspect that the OS monitors the fan speeds at the heatsinks and only kicks up to keep the heatsink at proper temperature. If you're not looking at core temps, the cores could be much hotter.

I used AS5 for the thermal compound so it's still "breaking in", but I did notice that the core temps have evened out now--they used to differ by 10C or more but now they're within a few degrees of each other. I'm probably going to pull everything apart again and replace the AS5 with noctua NT-H1--I used the AS5 because the procs arrived before the noctua and I couldn't wait! But it may not be worth the effort as I only expect a degree or two difference, unless my horizontal-line application technique of the AS5 wasn't "good".

Geekbench-32 (I need to pay those people!) went from 5207 stock to 9422 so in general all is as expected.

I don't have any definitive help for you because so far I've been blessed... but if it were me I'd suspect that my thermal compound job isn't good enough. Checking core temps with Hardware Monitor is important, you don't want to rely on the heatsink temp because if the thermal connection between CPU and heatsink is less than optimal the core can get quite hot in relation to the heatsink. And depending on the core temperature, it could cause problems... at least in my case high 70s were still stable.

Good to hear you've had such a smooth experience, enjoy the performance!

Taken your advice and re-applied my thermal compound to see if it makes any difference. At this stage temps much the same, Northbridge is always hotter than the core temps (currently at 63ºC whilst running default fan rpm while cores are between 40ºC and 51ºC). These measurements taken from Hardware Monitor as suggested.

Will keep tinkering...

Those temps are very much in line with what I'm seeing... at idle, default fan, 21C ambient

Something I noticed when I took the CPU heatsinks off the first time was that Apple applied their paste using what appeared to be a silk screened process. I make this assumption because it was visibly thick and had a fabric pattern in the areas that hadn't been squashed between the heatsink and the top of the CPU. Replacement thermal compounds instruct you to use very little, making the assumption that there wlll be zero gap between the top of the proc and the bottom of the heatsink when everything is torqued down and this will squish everything out uniformly to the edge of the processor-heatsink mating surface. I can get into why in general thin is better than thick, but the original thickness was surprising and worth noting.

I'm wondering if anyone knows if the design of the heatsink attachment is such that when fully torqued down, the heatsink isn't really in intimate contact with the surface of the heatsink? I think you'd notice this if you had applied your paste "as instructed" using a thin line or rice-grain sized blob without spreading it around, then attached the heatsink. Subsequent removal -should- have squashed that line/blob out completely over the surface of the CPU. If you remove the heatsinks and find that it's not squashed out completely, then there's probably a designed-in gap in the attachment mechanism, which may explain why Apple screened on such a thick layer of their paste.

If this is the case, then the technique for paste application on these machines may need to be altered!

I sure the repasting helps your KPs!

My experience with the original paste was the same. Also, when I removed the heatsinks a second time to reapply my thermal paste, I noticed that the paste had not really made contact with the heatsink as you have suggested. There was a small amount of contact on either side of the heatsink, probably about a 1mm strip on either side (no where near enough) so my application this time was more liberal that I would normally be comfortable with. I think your comments about the paste being screened and an intentional gap are spot on, I remember being surprised at how thick the past was originally, seemed much more like a pad than a thin smear of paste.