Mac Pro (Early 2008) Geekbench Benchmarks

[MacRumors]

PrimateLabs provides some early Geekbench 2 benchmarks on the just-released 8-Core "Harpertown" Mac Pro at 2.8GHz compared to the old 8-Core "Clovertown" 3.0GHz Mac Pro.

Despite the slower clock speed, the new "Hapertown" Mac Pro outperformed the older "Clovertown" Mac Pro in Integer, Memory and Stream performance, reflecting its faster memory (800MHz vs 667MHz) and bus (1600MHz vs 1333MHz). The new Mac Pro, however, did lag behind in floating point performance, compared to the older higher-clock speed model.

Apple introduced the new Mac Pro earlier this week.

Article Link

[WildPalms]

Dazzling! Ahh yes, its good to be Mac, dont wanna go back!

[Eidorian]

I believe that the biggest benefits will be seen in SSE4 applications. Video editing sees big improvements in this department.

On another front 45nm is a heck of a lot cooler as well. I've seen the heatsinks for the new 45nm Core 2 Duo and they chopped it in half from the older 65nm Core 2 Duo!

Intel seems confident that these new processors are cool if they're doing that to their somewhat respectable stock fans.

[eXan]

I think that real-world benchmarks would be more useful than this synthetic test...

[Plumbstone]

Pretty pleased that my year old 8-core MacPro still wins (just) on overall score with half the RAM...Was a bit more expensive though so £ for £ the new ones are a pretty good deal....

[mmendoza27]

Of course, when apple releases an update to take advantage of the SSE4 instruction set, we'll see updates to these benchmarks and we should see some hefty improvements.

I think Apple did a good job updating an already beautiful workstation.

[Eidorian]

Originally Posted by mmendoza27

Of course, when apple releases an update to take advantage of the SSE4 instruction set, we'll see updates to these benchmarks and we should see some hefty improvements.

I think Apple did a good job updating an already beautiful workstation.

Rumor goes that 10.5.2 is Penryn enhanced. It was thought that this release was holding back the Mac Pros.

[janvandervoort]

when you consider the fact that the new Mac Pro is so much more affordable than the old Mac Pro and it’s hard to call this upgrade anything less than impressive

For those who are not pleased by the fact that the box wasn't upgraded: the MacPro is a classic design that will never ever feel dated...

[netdog]

Originally Posted by Eidorian

I believe that the biggest benefits will be seen in SSE4 applications. Video editing sees big improvements in this department.

On another front 45nm is a heck of a lot cooler as well. I've seen the heatsinks for the new 45nm Core 2 Duo and they chopped it in half from the older 65nm Core 2 Duo!

Intel seems confident that these new processors are cool if they're doing that to their somewhat respectable stock fans.

Yes, SSE4 is not even implemented yet at the OS-level, though I think we all expect that to change on Tuesday.

Will Geekbench need a rewrite to be able to correctly identify those gains or will the OS handle it.

In either case, given that SSE4 makes for MAJOR speed increases according to Intel, given these results today, when rendering we can expect the 2.8 to...dare I say it...blow away the old 3.0 octo. Frankly, video is the main reason I got this machine (or rather that my wonderful woman got it for me) and I am really impressed given what I have read.

I will be able to speak from real-world experience once the RAM and drives arrive!

[richthomas]

what about this one? http://browse.geekbench.ca/geekbench2/view/32012 this person has a score of 13992 using a 1.83 GHz core duo with 2 gigs memory and Mac Pro mobo. is this for realyo?

[netdog]

Originally Posted by richthomas

what about this one? http://browse.geekbench.ca/geekbench2/view/32012 this person has a score of 13992 using a 1.83 GHz core duo with 2 gigs memory and Mac Pro mobo. is this for realyo?

Makes you wonder why we bought Mac Pros, eh?

[Eidorian]

Originally Posted by netdog

Yes, SSE4 is not even implemented yet at the OS-level, though I think we all expect that to change on Tuesday.

Will Geekbench need a rewrite to be able to correctly identify those gains or will the OS handle it.

Geekbench needs to be re-written in order to show the gains from newer instruction sets and CPU features.

I doubt that they will though. Geekbench is designed to be a generic and broader benchmark for a variety of processors. It's "supposed" to truly benchmark a processor instead of taking advantage of each processors special little tricks to make it faster.

[budward]

Okay, I have a Mac Pro 2.66 with 4 x 1GB sticks...

I just bought a new Mac Pro from Apple and didn't choose for Apple to upgrade my RAM (you guys know why) so It comes with 2 x 1GB ..

If I was to buy 2 x 2gb from OWC can I mix 1GB and 2GB sticks? Or am I going to run into problems with? That would give me a total of 6GB RAM

1GB x 2
2GB x 2

I guess I am asking if I need to stay consistent with the same size ram throughout.. and if not does it effect the Mac Pro in anyway.. (It's all 800MHZ) not the older Mac Pro ram...

Thanks in advance.

[fluidedge]

nice

i'd be very interested to know about the new MP 'single 2.8 GHz chip' benchmarks

[richthomas]

Originally Posted by netdog

Makes you wonder why we bought Mac Pros, eh?

i think it may be a bug in geekbench or geekbench not liking hackintosh's non Apple liscenced hardware. i saw a few other TGTBT scores but im just waiting to hear if its true.

[allpar]

...these new processors into existing Pros?

[nickane]

Since the april 8-core came at such a premium, I guess the many shoestring professionals who waited for 8-core across the board can be pleased they saved their money all this time...

[gnasher729]

Originally Posted by allpar

...these new processors into existing Pros?

Maybe, but the fast versions of the chips are rather expensive; you would probably be better off financially selling an old MacPro on eBay and buying a new one instead of trying to upgrade it.

There are other risks apart from the obvious one that you might damage the Mac or the processors: It could be that old MacPro's "know" that they have a chip without SSE4. An application that uses SSE4 needs to check whether it is available, because otherwise it would just crash. Now if an app does that and the computer says "no SSE4 here" even though there is a Penryn chip there, then you won't get any improvements coming from new code, only improvements from the fact that old code still runs faster.

[BenRoethig]

Originally Posted by netdog

Yes, SSE4 is not even implemented yet at the OS-level, though I think we all expect that to change on Tuesday.

Will Geekbench need a rewrite to be able to correctly identify those gains or will the OS handle it.

In either case, given that SSE4 makes for MAJOR speed increases according to Intel, given these results today, when rendering we can expect the 2.8 to...dare I say it...blow away the old 3.0 octo. Frankly, video is the main reason I got this machine (or rather that my wonderful woman got it for me) and I am really impressed given what I have read.

I will be able to speak from real-world experience once the RAM and drives arrive!

Yeah, I would expect full SSE4 support before too long. Rumor has it, SSE4 was designed with Apple in mind.

[Full of Win]

News Flash: Two computers with different amounts of RAM, different GHz's and different FSB's preformed differently on test!

One of the most basic tenets of experimental design is to remove as many of the variables as possible before running any test. The fewer variables, then the easier it is to attribute a reason to what remains as different. There are way to many points of difference to make much use out of these numbers.

[Small White Car]

Originally Posted by budward

If I was to buy 2 x 2gb from OWC can I mix 1GB and 2GB sticks?

The computer and the OS support what you're suggesting. It will work.

HOWEVER...
The Final Cut Pro documents strongly suggest keeping everything symetrical:

http://docs.info.apple.com/article.html?artnum=304492

So you have to ask yourself, if Apple's high-end software suggests this, might it be something I should be doing, even if I'm not using Final Cut Pro?

[heisetax]

Originally Posted by budward

Okay, I have a Mac Pro 2.66 with 4 x 1GB sticks...

I just bought a new Mac Pro from Apple and didn't choose for Apple to upgrade my RAM (you guys know why) so It comes with 2 x 1GB ..

If I was to buy 2 x 2gb from OWC can I mix 1GB and 2GB sticks? Or am I going to run into problems with? That would give me a total of 6GB RAM

1GB x 2
2GB x 2

I guess I am asking if I need to stay consistent with the same size ram throughout.. and if not does it effect the Mac Pro in anyway.. (It's all 800MHZ) not the older Mac Pro ram...

Thanks in advance.

The only restriction is that the dimms be put in with exact pairs, ie same size, speed, numbers of memory chips & such. The inside computer covers shows how to put your dimms in. In my 17 month old Intel Mac Pro I have 6 2 GB dimms & the orginal 2 512 MB dims for a total of 13 GB.

I'm sure that you will have a good time with your new Intel Mac Pro. I have with mine.

Bill the TaxMan

[guzhogi]

I posted this in the other Mac Pro thread on the front page that a good place to look for RAM is dealram.com. It has a comparison of prices for kinds of RAM. You can chose specific kinds of RAM (PC3200, PC4200) or specific computer models (Apple Mac Pro, iBook, or Dell and Gateway computers). You can also get flash memory (Compact Flash, SD, even USB flash drives). I was able to get a 32 GB (yes, gigabyte) USB flash drive for only $235. Only thing is I don't think it lists the new Mac Pro memory yet.

[Raid]

Originally Posted by Full of Win

News Flash: Two computers with different amounts of RAM, different GHz's and different FSB's preformed differently on test!

Quote for truth! I guess though the intent was to see how one of the new ones stacked up to the first 8-core Mac Pro.

A much better test would be comparing an 8-Core 3.0Ghz with 2 GB RAM to a similar spec'd Early 2008 Mac Pro. That would truly show off the change in performance of old vs. new.

[jragosta]

Originally Posted by eXan

I think that real-world benchmarks would be more useful than this synthetic test...

Agreed. Even more importantly, they need to do more to minimize variables. It would be nice if they could have used the same clock speed on both machines.

Even if that weren't possible, why didn't they put the same amount of RAM in both machines? Twice the RAM in the newer machine could be skewing the results quite a bit. Equalizing the RAM is an easy thing to do.