Benchmarks for all three 8-Core Penryn Mac Pros reported
Following up on a previous set of 2.8GHz Mac Pro benchmarks, Geekbench developer Primate Labs has expanded its report on the latest Apple workstations to include performance measures for all three models, including the 8-Core 3.2GHz configuration.
Each of the three machines were configured identically with 2.00GB of 800 MHz DDR2 FB-DIMMs, Mac OS X 10.5.1 build 9B2117, and pit against a baseline, where a score of 1000 represents the performance of a Power Mac G5 running at 1.6GHz.
Additionally, each model was tested under Geekbench 2 running both 32-bit and 64-bit code for a more in-depth comparison of the systems' capabilities.
"What?s interesting about the charts [below] is that the performance difference between the 2.8GHz and 3.2GHz Mac Pro isn?t as great as the difference between running 32-bit code and 64-bit code," the firm concluded following the tests. "In fact, the 2.8GHz Mac Pro running 64-bit code is faster than the 3.2GHz Mac Pro running 32-bit code."
Each of the three machines were configured identically with 2.00GB of 800 MHz DDR2 FB-DIMMs, Mac OS X 10.5.1 build 9B2117, and pit against a baseline, where a score of 1000 represents the performance of a Power Mac G5 running at 1.6GHz.
Additionally, each model was tested under Geekbench 2 running both 32-bit and 64-bit code for a more in-depth comparison of the systems' capabilities.
"What?s interesting about the charts [below] is that the performance difference between the 2.8GHz and 3.2GHz Mac Pro isn?t as great as the difference between running 32-bit code and 64-bit code," the firm concluded following the tests. "In fact, the 2.8GHz Mac Pro running 64-bit code is faster than the 3.2GHz Mac Pro running 32-bit code."
Comments
Penryn? Penryn Processors are 45 nm Core2 processors. The Xeons the Mac Pro are using now are codenamed "Harpertown"
I thought someone explained that Penryn is the codename for the newest class of CPUs, and that Harpertown is one of the Penryn CPUs.
I thought someone explained that Penryn is the codename for the newest class of CPUs, and that Harpertown is one of the Penryn CPUs.
Yeah, I think you are right, it appears to be referring to ALL of the 45nm ones. Sorry.
..and make myself less depressed about not being able to afford a Mac Pro.
-Clive
Why would most of the performance data for the 3.0GHz not be in the middle between the 2.8 and 3.2? Nearly all 3.0GHz data is close to the 2.8GHz numbers, and for memory throughput the numbers for the 3.0GHz are actually worse than the 2.8GHz.
Either they got a bad unit, or they got their configurations messed up. Pity really, was looking at getting the 3.0GHz, now I've got some FUD!
Can someone pit a Core 2 Duo Mac Mini against a first gen 800MHz iMac G4? I want to psych myself up for my next purchase...
..and make myself less depressed about not being able to afford a Mac Pro.
-Clive
http://www.barefeats.com/mincd.html
1.42 G4 Mini vs 1.83 Core Duo Mini.
Should be close enough for your purposes. Hope you don't play games...I really wish the mini would get the GMA X3100.
Interesting result from the 32 versus 64 bit...I've seen tons of FUD from people who are adamant that apps shouldn't be updated to 64 bit code because they think it will cut performance in half.
Not in half, but moving to 64 bits can significantly decrease an applications performance, all things being equal. If we'd stayed on PowerPC, moving to 64-bits probably would have decreased performance. However, on the x86 instruction set the move from 32 bits to 64 bits doesn't just change the register file's width. They also cleaned up the ISA, doubled the number of registers, and streamlined the design, so it's not just about the bits.
In addition, specific operations do benefit from 64-bits, including ones that require more than 4GB of RAM, ones that deal in numbers higher than 2^32 regularly, ones that process bitstreams in the CPU instead of SIMD or GPU, and a few others. Browsing a web or using MS Word or your "typical" desktop app doesn't benefit from the extra bits (but it still benefits from the ISA cleanup noted above on x86.)
I havn't seen any benchmarks for the single processor Pro. Actually, I have yet to talk to anyone who has purchased one. On a budget, the single processor pro seems like a sweet deal compared to the 2.8ghz extreme iMac.... but is it worth it???? Should I just save up the extra $$ and get the baseline 8 core?
Just looking for some opinions, I just sold my iMac and am looking at the new single cpu pro. I use logic pro a few hours a day and need some more juice! I don't run any external plug-ins or CPU intensive synths in Logic, so the 8 core isn't really necessary.
I havn't seen any benchmarks for the single processor Pro. Actually, I have yet to talk to anyone who has purchased one. On a budget, the single processor pro seems like a sweet deal compared to the 2.8ghz extreme iMac.... but is it worth it???? Should I just save up the extra $$ and get the baseline 8 core?
the 2 socket 8-core 2.8GHz mac pro is the new baseline, there are no single cpu/socket options anymore. Maybe you could get one of the older mac pro's at a good price now if you're quick and don't want the latest/greatest.
the 2 socket 8-core 2.8GHz mac pro is the new baseline, there are no single cpu/socket options anymore. Maybe you could get one of the older mac pro's at a good price now if you're quick and don't want the latest/greatest.
You can get a single 2.8 at the build to order page. Hit "configure now" at the Apple store page and you should see as the top option: "One 2.8GHz Quad-Core Intel Xeon (quad-core) [Subtract $500]"
The old dual 3.0 may be as good of an option, if you can get it for cheaper, but $2299 is a pretty good price for a minimum Mac Pro. I wouldn't have considered the old minimum of 2.0 Quad unless bandwidth was a huge consideration.
the 2 socket 8-core 2.8GHz mac pro is the new baseline, there are no single cpu/socket options anymore. Maybe you could get one of the older mac pro's at a good price now if you're quick and don't want the latest/greatest.
You can buy a Mac Pro with only one cpu and presumably the other socket ready for another cpu if you wish.
See here.
http://www.barefeats.com/mincd.html
1.42 G4 Mini vs 1.83 Core Duo Mini.
Should be close enough for your purposes. Hope you don't play games...I really wish the mini would get the GMA X3100.
Wow, pwnage! Thanks!
Yeah, I definitely hear you about the GMA X3100. I actually do plan on waiting for S.R. on the Mini, so I'll have my cake and eat it too.
-Clive
the 2 socket 8-core 2.8GHz mac pro is the new baseline, there are no single cpu/socket options anymore. Maybe you could get one of the older mac pro's at a good price now if you're quick and don't want the latest/greatest.
???????
http://store.apple.com/1-800-MY-APPL...QWPbW2Y/2.?p=0
\t
Processor
Incredible 8-core performance, up to 3.2GHz, is available with two Quad-Core Intel Xeon ?Harpertown? processors. Or configure a quad-core system consisting of one 2.8GHz Quad-Core Intel Xeon processor.
\tOne 2.8GHz Quad-Core Intel Xeon (quad-core) [Subtract $500]
\tTwo 2.8GHz Quad-Core Intel Xeon (8-core)
\tTwo 3.0GHz Quad-Core Intel Xeon (8-core) [Add $800]
\tTwo 3.2GHz Quad-Core Intel Xeon (8-core) [Add $1600]
Looks like they got a dud 3.0GHz unit.
Why would most of the performance data for the 3.0GHz not be in the middle between the 2.8 and 3.2? Nearly all 3.0GHz data is close to the 2.8GHz numbers, and for memory throughput the numbers for the 3.0GHz are actually worse than the 2.8GHz.
Either they got a bad unit, or they got their configurations messed up. Pity really, was looking at getting the 3.0GHz, now I've got some FUD!
It is't actually that unusual when two otherwise identical machines with processors that are so close in speed reverse specs in some areas.
The cpu's may not be running at exactly the speed issues. Memory may be a bit fadter, or slower. The bus may have tiny glitches, etc. It's just a few percent.
By the way, regarding speed, BareFeats has an interesting memory test.
They found that the new FB 800 Penyrn machines WILL work with the older 667 memory, and that, on the machine they tested, you only lose between 1 and 4% performance.
While they aren't recommending that you use them permanently, they are saying that you can transfer the extra's from the older machine until you get more fastr memory.
Also, as was true for the older machines, the memory buss performs better if the first two memory slots from each board are filled.
In other words, if you have four Gb of RAM, it's faster to have it in 4 1GB sticks, than in 2 2GB sticks.
Lastly, if you have 2GB of Apple RAM at 1GB per stick and then get 8 GB in 2GB sticks, put the 2GB Apple RAM in the TOP card with the 2 2GB sticks, and put the other 2 2GB sticks in the first two slots of the bottom card.
THose still deciding, you may want to check out BareFeats testing. The info on RAM was enlightening.
I'm still waiting to see how the 2.8GHz Harpertown compares to the 2.66GHz Woodcrest and 3.0 Clovertown.
edit: pipped by Melgross... again!
Not in half, but moving to 64 bits can significantly decrease an applications performance, all things being equal. If we'd stayed on PowerPC, moving to 64-bits probably would have decreased performance. However, on the x86 instruction set the move from 32 bits to 64 bits doesn't just change the register file's width. They also cleaned up the ISA, doubled the number of registers, and streamlined the design, so it's not just about the bits.
In addition, specific operations do benefit from 64-bits, including ones that require more than 4GB of RAM, ones that deal in numbers higher than 2^32 regularly, ones that process bitstreams in the CPU instead of SIMD or GPU, and a few others. Browsing a web or using MS Word or your "typical" desktop app doesn't benefit from the extra bits (but it still benefits from the ISA cleanup noted above on x86.)
I guess I should clarify. I've heard FUD from people who think that even apps that do heavy number crunching and could benefit from more than 4 gigs of ram shouldn't be updated, because there will be a huge performance hit. I have yet to see any real data backing that up, just seems to be paranoid speculation. Obviously, it would be silly to update Textedit to 64 bit, but the notion that every app would take a performance hit seems pretty ridiculous.
It's not a dud. The 3.0GHz Octo uses the old generation CPU, its the model that's been replaced. Its quad core chips were two dual core chips sewed together by the front side bus. The new chips have all four cores on one die, I think with 50% more cache, 16% faster memory and CPU bus, a fully shared cache pool and new instructions and optimizations. The new systems may be more power efficient than the old octo, I don't know.
That's not what the article says. All three models are listed as "early 2008". I have no idea why the 3.0 would lag the 2.8 on any tests, that does seem fishy.