Just upgraded Mac Pro to 28 core from 8 core, easy peasy

For those considering a CPU upgrade I just did it with little to no fanfare.
I bought my base 8 core with 48 gigs of ram Mac Pro from the Apple refurb store with the thought of upgrading the CPU down the line after the graphics card and of course ram.
But I came across some OEM Intel Xeon W-3275M processors on -Bay last week for $2,999 so I spent way to much time researching for a few days then pulled the trigger.
This is the exact 28 core cpu Apple puts in the Mac Pro and so far it acts that way.

There was apparently an issue when swapping the CPU in MacOS which was fixed in 11.2 Big Sur.
I put in it, plopped on some thermal paste, closed it up, powered it on and so far smooth sailing.
I currently have 240gigs of ram and when using the ML_timewarp I can get all 56 cores cooking on say a 2k plate.

The seller is a legit eBay computer parts store and there are 4 left. If you search for Intel Xeon W-3275M in -Bay or NewEgg it’ll come up as this is/seems to be thinly ones for sale at all currently.
It’s the same seller on NewEgg but actually costs a little more.


Awesome. Thanks Ben!

1 Like

Have you benchmarked any of your comps? I see flame stuck at 99% / 2000% (ten core) so often that I wonder if it’s worth it to go to a proc with slower clock speed.

1 Like

Sam, ya got a test setup you want to send me? This weekend I could benchmark a 16vs28 for you.

1 Like

Sam I haven’t yet, it definitly beats my old trashcan:)
Would be interesting to see with Randy’s two machines for sure.

One thing about general speed I learned during my research is that with these CPU’s the real world speed when not being fully multi-threaded is the Turbo boost speed, not the base clock.
I found this is easier to see looking at Intel Power Gadget vs iStats Pro. Seems like because iStats is just using like a mean average or something it’s not showing all the times cores top out at say 4.4 for the 28 core.
The CPU will select the better performing cores at the moment it’s called for use. When things start really kicking in then it starts to balance the available power vs the needed cores.
At least that’s now my understanding of it, but I’m def not computer scientist/mathmatician chip engineer… LOL