WARNING: THIS IS GOING TO BE A LONG POST
I recently performed a CPU upgrade on the 2006 Mac Pro 1,1 I got off craigslist for 150 dollars.
So I picked up a 2006 Mac Pro for 150 bucks. Has no GPU, but I have arranged to get a 7950 Mac edition for 80 bucks (which is a complete and utter loving score, prices are currently 270 and up still) and it is technically been re-branded to an R9 280 in relative performance.
I as well placed a reference HD 7950 in it that I got off craigslist for 80 dollars
So I managed to score some sweet stuff on craigslist tonight. Got an HD 7950 3Gb card for 80 dollars and I even managed to sell off at the same time my Galaxy Tab 3 for 150 bucks. Now to just upgrade the processors on the cheese grater mac pro I bought and it'll be rocking.
Later on I put 20Gbs of DDR2 667 Mhz ECC RAM in which can be topped out at 32Gbs if I so wanted to later on.
In 2006, which the Mac Pro 1,1 came out, the model I obtained was the 2.66 Ghz 2X Xeon 5150 tier. This specific model retailed at roughly $2499 USD back in August of 2006, however with the CPU upgrade of 2X Xeon X5355's, a set of processors that retailed at $1172 (per chip) which would total this machine at $4,843 USD before taxes or other fees. After performing some Passmark tests on both the 5150's and the X5355's, it is quite evident that even with processors that were manufactured in 2005-2006, can still perform just as fast, if not in some cases, faster than modern day processors. *Note* some scenarios for gaming the Mac Pro 1,1 does not handle the best with PCIE 1.0 lanes halving the bandwidth*
I noted on GTA V, on ver high and 8X AA, all advanced graphics on, 60 fps was easily achievable in the intro of the game, however, some explosions and other high activity scenes diminished to 40 fps. This is more than likely due to PCIE 1.0 having a bottleneck on the 7950 performing to it's best ability versus a CPU bottleneck.
PASSMARK RESULTSIn these tests I compared the CPU results of both xeons against my existing 6350 OC'd at 4.20 Ghz on a consistent basis while being watercooled by a Corsair H105.
In summary, I note that the X5355's improved twice over the 5150's while still lagging slightly behind the 6350, which has the benefit of being nearly a full 2 Ghz faster in clock speed over the 2.66 X5355's. The X5355's however do maintain a strength over many processors today with the elimination of L3 cache, and focusing on a large L2 cache which is why the prescence of L3 cache even exists. L3 cache is stored off chip allowing for larger storage, but being off chip requires more time to access set instructions for the CPU, with L2 being large at a beefy 8 Mb per chip on the X5355's, it doubles up to become 16 Mb of L2 cache achieveing much faster memory access within the chip's themselves.
I found out that the FX-6350 has a slightly unbalanced advantage being nearly 2 GHz faster, however, we are comparing two 2005 processors to a 2013 processor. Looking at our first column, we see the 6350 at 7197, the 5150 at 3677 and finally our new X5355’s at 6308. Giving us a 2600 point increase, doubling what it was before, even though it still has not beaten the 6350. When it came to compression measured at Kbytes per second, the X5355’s truly outshined both the 6350 and the 5150’s at 12048 placing it 4.9% faster than an average result of a FX-8150 leaving it still 60.1% behind a i7-5820K at 3.30 GHz.
Single threading is a category I feel doesn’t exemplify the Xeon’s in any light, this is due to the fact Xeon’s are built to dual process (spreading out a work load between two CPU chips) and often not will single thread on a single chip, so I just look at this category as exempt. Extended Instructions (SSE) showed a large improvement on the X5355’s over the 5150’s at an improvement of 10.5 Million Matrices per second. This left the X5355’s at 6.6 behind the 6350 and shockingly, the Xeon’s were less in physics at 352.8 frames per second where the 6350 was at 459.3 frames per second. I would have presumed the X5355’s to speed ahead on this category due to the fact most physics powered games and applications are directly CPU and with dual processing and hyper threading (however, the X5355's do not have hyper threading technology due to age) it should triumph easily over the 6350 and it’s glorified 3 physical cores and 6 logical versus the X5355’s 8 physical cores and 8 logical cores.
TL;DR
A 2006 Mac Pro 1,1 with 20 Gbs of RAM and two quad core processors with $339.98 USD put into this thing total, makes one hell of a machine in comparison to some machines in 2015. Yes, it is slower in terms of RAM and graphics bandwidth, but if it needs to run anything CPU intensive, this thing will knock and deliver results.
Bonus Pictures
Poll