Primate benchmarks Apple's new 8-core Mac Pro

[AppleInsider]

Just days after its introduction, the folks at Primate Labs have grabbed hold of one of Apple's new 2.8GHz Harpertown-based Mac Pros and pit the system against its 3.0GHz Clovertown-based predecessor in a set Geekbench benchmark tests.

The results saw the new Mac Pro with its two 45-nanomete (nm) quad-core 2.8GHz chips and 2GB of 800MHz DDR2 RAM nearly match the overall performance of the previous-generation Mac Pro, which employed two 65 nm quad-core 3.0GHz chips and 1GB of 667MHz DDR2 RAM.

While the Harpertown Mac Pro trailed the Clovertown machine in floating point performance, it bested the system in integer, memory and stream performance. Overall, the Hapertown Mac Pro achieved a Geekbench 2 score of 7598 compared to the Clovertown Mac Pro's 7680.

The results are impressive, notes Primate, when you consider that the new Mac Pro is much more affordable than the old Mac Pro. Apple also offers the latest Mac Pro at speeds of up to 3.2GHz -- a configuration which should handily trounce the 3.0GHz Clovertown machine in each and every benchmark.

[ View this article at AppleInsider.com ]

[solipsism]

How is this "much more affordable" given the $300 increase for the 2nd model, and the base model being the same price but only having one processor?

edit: I see. They are comparing the suggested model with the previous high-end machine.

[mikenap]

I thought i had read that the 2.8 was a different chip than the 3.0 and 3.2, with a slower frontside buss? Anyone know for sure?

[jragosta]

It would be nice if they'd make an effort to minimize the variables. In this comparison, they've changed the clock speed, chip type, and amount of RAM. Ideally, they would have compared 3.0 GHz Cloverton to 3.0 GHz Penryn. Even if they didn't have the 3.0 GHz Cloverton, they should have AT LEAST put the same amount of RAM in both machines. It's impossible to tell how much the RAM difference affected the results, so how do you draw conclusions?

[solipsism]

Originally Posted by mikenap

I thought i had read that the 2.8 was a different chip than the 3.0 and 3.2, with a slower frontside buss? Anyone know for sure?

Processing
8-core: Two 2.8GHz, 3.0GHz, or 3.2GHz Quad-Core Intel Xeon 5400 series processors

Quad-core: One 2.8GHz Quad-Core Intel Xeon 5400 series processor

Enhanced Intel Core microarchitecture
• 12MB of L2 cache per processor (each pair of cores shares 6MB)
• 128-bit SSE4 SIMD engine
• 64-bit data paths and registers
• Energy efficiency optimization

1600MHz, 64-bit dual independent frontside buses

(source)

They all use the same high-end Xeon L2 cache and FSB.

[roehlstation]

Originally Posted by jragosta

Ideally, they would have compared 3.0 GHz Cloverton to 3.0 GHz Penryn.

Did I completely not get what you were saying? How would it be ideal to compare the Clovertown to the Penryn? They don't use the Penryn in the Mac Pro, it is a Core 2 processor not a Xeon.

[suhail]

Yawn!
Arguably the old and the new have the same performance, a slightly different price tag, and probably lower power consumption for the 45 nm.
Yeah a definite Yawn.

For many, speed is not an issue anymore but size and price are. Hopefully we won't be disappointed on Tuesday and we'd see an xMac with at-least one PCI slot.

[solipsism]

Originally Posted by roehlstation

Did I completely not get what you were saying? How would it be ideal to compare the Clovertown to the Penryn? They don't use the Penryn in the Mac Pro, it is a Core 2 processor not a Xeon.

The 5400 series Intel Xeon used in the Mac Pro are Intel Core 2 "Harpertown" Penryns.

Originally Posted by suhail

Arguably the old and the new have the same performance, a slightly different price tag, and probably lower power consumption for the 45 nm.

They are testing the baseline 8-core setup with the previous model's top-of-the-line setup which makes the price difference very significant.

Apple doesn't list of the older 8-core systems but they do have the 4-core systems @ 3.0GHz and 2GB RAM for $3,300. I'm guessing the older 8-core systems would add at cool $1000 on top of that.

As for performance, it would have been nice if they used the same amount of RAM, but they processor tests show a very good machine for the price, not to mention that once Leopard can utilize SSE4 there will be a nice jump in performance. This makes this new machine a future-proof investment.

[zanshin]

it appears they compared the previous top-of-the-line dual processor Mac Pro with the new bottom of the line DP model, both in stock config as sold by Apple.

The new one low-end DP Mac Pro (stock config) comes with more RAM, better video, enhanced card slotting and memory, larger HD, and the new Al keyboard, for substantially less money, and closely matches performance with the previous top-tier model. And unlike 8-core Mac Pros bought before October 07, it includes Leopard. Also has more upfront ports, I believe.

It seems pretty apparent that it's an overall better deal.

I can't think of a single thing I do on a Mac where I wouldn't appreciate more speed. Faster app loading, better performance, less beach-balls. I know I'd get more out of typing better, but overall, more speed is never a bad thing

[solipsism]

Originally Posted by zanshin

it appears they compared the previous top-of-the-line dual processor Mac Pro with the new bottom of the line DP model, both in stock config as sold by Apple.

The new one low-end DP Mac Pro (stock config) comes with more RAM, better video, enhanced card slotting and memory, larger HD, and the new Al keyboard, for substantially less money, and closely matches performance with the previous top-tier model. And unlike 8-core Mac Pros bought before October 07, it includes Leopard. Also has more upfront ports, I believe.

It seems pretty apparent that it's an overall better deal.

I can't think of a single thing I do on a Mac where I wouldn't appreciate more speed. Faster app loading, better performance, less beach-balls. I know I'd get more out of typing better, but overall, more speed is never a bad thing

front ports look the same to me. Don't forget it also comes with Bluetooth.

PS: Anyone know how much the previous 8-core model cost new?

[r00fus]

Originally Posted by jragosta

It would be nice if they'd make an effort to minimize the variables. In this comparison, they've changed the clock speed, chip type, and amount of RAM. Ideally, they would have compared 3.0 GHz Cloverton to 3.0 GHz Penryn. Even if they didn't have the 3.0 GHz Cloverton, they should have AT LEAST put the same amount of RAM in both machines. It's impossible to tell how much the RAM difference affected the results, so how do you draw conclusions?

Yeah, it's not really a benchmark, but more like a "basemark"... it's the lowest priced model (though the price is different) despite the processor speed, memory and FSB speed differences.

All it really says is, if you were going to buy the BASE model Mac Pro, you get more for your dollar, and you don't have to pay as much.

I'd really have preferred a much more direct comparison, esp. with the RAM (ie, 2GB vs 2GB)

[digitalclips]

Does any one else think these new G5s may be blow out models clearing the cases and motherboards in advance of something totally new at WWDC? Or am I too cynical ...?

[jfpoole]

Originally Posted by jragosta

It would be nice if they'd make an effort to minimize the variables. In this comparison, they've changed the clock speed, chip type, and amount of RAM. Ideally, they would have compared 3.0 GHz Cloverton to 3.0 GHz Penryn. Even if they didn't have the 3.0 GHz Cloverton, they should have AT LEAST put the same amount of RAM in both machines. It's impossible to tell how much the RAM difference affected the results, so how do you draw conclusions?

Would you have preferred a comparison between two identical machines? That would certainly reduce the variables! It'd just make the comparison completely useless.

The point of this comparison was to compare the new standard Mac Pro against the old high-end Mac Pro and take a look at the differences. I'm not sure what benefit using 3.0GHz CPUs would have been except increase the lead the new has over the old.

Oh, and there's an explanation in the article as to why the old Mac Pro had less RAM than the new Mac Pro:

It might seem unfair to compare a new Mac Pro with more RAM against an old Mac Pro with less RAM. However, both Mac Pros have two FB-DIMMs installed. Since Mac Pro performance increases dramatically when it’s configured with four FB-DIMMs sticks, I figure it’s more important to make sure the two Mac Pros have the same number of FB-DIMMs installed rather than the same amount of memory installed.

Also, Geekbench itself isn’t incredibly memory hungry (Geekbench prefers faster RAM instead of more RAM) so I doubt the performance of the old Mac Pro will suffer from having less memory installed.

[solipsism]

Originally Posted by digitalclips

Does any one else think these new G5s may be blow out models clearing the cases and motherboards in advance of something totally new at WWDC? Or am I too cynical ...?

I believe Melgross has made mention that next Intel chip will warrant a new exterior case design, but I don't think so. The new one still looks great and I find it hard to conceive of how they could improve on this timeless design. Now, if the heat and innards are reduced then perhaps it would warrant some changes to fully optimize the space, but I don't think it will. Just my opinion.

PS: These aren't G5.

Originally Posted by jfpoole

Would you have preferred a comparison between two identical machines? That would certainly reduce the variables! It'd just make the comparison completely useless.

The point of this comparison was to compare the new standard Mac Pro against the old high-end Mac Pro and take a look at the differences. I'm not sure what benefit using 3.0GHz CPUs would have been except increase the lead the new has over the old.

Oh, and there's an explanation in the article as to why the old Mac Pro had less RAM than the new Mac Pro:

I guess we could have it both ways. One to see a straight comparison of clockspeed and GB of RAM and another test to see how the new base model "Harpertown" 8-core compares to the previous "Clovertown" 8-core.

I did find that the machine tested would have cost $3,997 and not include Bluetooth, a 320GB HDD and an extra 1GB RAM. Anyone who bought a Mac Pro before Tuesday and within the last two weeks should return it.

[tonton]

Twice as many stamps on this one...

[tonton]

This is the funniest headline EVER.

But was it this technician...



...or this one?

[melgross]

Originally Posted by tonton

This is the funniest headline EVER.

We all knew those lab guys were monkeys.

Damn!

You beat me to it!

I read through all of the posts to see if anyone caught it, and it had to be the last one.

[melgross]

Originally Posted by tonton

This is the funniest headline EVER.

But was it this technician...



...or this one?

It could also be Ballmer. He's always referred to as "Monkey Boy" ever since his onstage performance.

[melgross]

I'm not sure what they mean the "floating point performance is off the charts."

It certainly looks as though the Clovertown has this one beat by a good amount.

As has been mentioned, equal amounts of memory should have been used.

While I often agree that supplied memory should be compared for consumer machine tests, because few consumers get more memory, with pro machines, that is not often the case.

2GB RAM is really sufficient for most pros even. But 1GB isn't.

I suppose we'll just have to wait for someone else to get one, such as Barefeats. If fact, I'll check now! Be back soon.

Tick

Tick

Tick...

Ok, They don't have one yet, but they do have some info, and interesting advice;

http://www.barefeats.com/index.html

[jragosta]

Originally Posted by roehlstation

Did I completely not get what you were saying? How would it be ideal to compare the Clovertown to the Penryn? They don't use the Penryn in the Mac Pro, it is a Core 2 processor not a Xeon.

The current Harpertown chips are part of the Penryn family.

Perhaps it would have been clearer if I had said that they should have compared 3.0 GHz Harpertowns to the 3.0 GHz Clovertown, but the essence of my statement is the same.

[jragosta]

Originally Posted by jfpoole

Would you have preferred a comparison between two identical machines? That would certainly reduce the variables! It'd just make the comparison completely useless.

The point of this comparison was to compare the new standard Mac Pro against the old high-end Mac Pro and take a look at the differences. I'm not sure what benefit using 3.0GHz CPUs would have been except increase the lead the new has over the old.

Oh, and there's an explanation in the article as to why the old Mac Pro had less RAM than the new Mac Pro:

Obviously you don't have ALL the variables the same. But good science involves reducing the variables as much as possible.

As I said, it MIGHT be justifiable to compare the different clock speeds. But their explanation for using different RAM levels is silly. It would be far more meaningful to use 2 1 GB DIMMs in each machine.

[jragosta]

Originally Posted by digitalclips

Does any one else think these new G5s may be blow out models clearing the cases and motherboards in advance of something totally new at WWDC? Or am I too cynical ...?

No. That would be insanely stupid.

(BTW, Apple hasn't sold G5s for quite a while. I'm assuming you meant Intel chips.)

[Ronbo]

Originally Posted by suhail

Hopefully we won't be disappointed on Tuesday and we'd see an xMac with at-least one PCI slot.

Man, you're just all set up for being disappointed until the end up time, aren't you?



(Seriously, though I'm sure there's a market for what you're pining for, if Apple wanted to pursue that niche, they'd have surely done it long ago, right?)

[digitalclips]

Originally Posted by solipsism

I believe Melgross has made mention that next Intel chip will warrant a new exterior case design, but I don't think so. The new one still looks great and I find it hard to conceive of how they could improve on this timeless design. Now, if the heat and innards and reduced then perhaps it would warrant more some changes to fully optimize the space, but I don't think it will. Just my opinion.

PS: These aren't G5.


I guess we could have it both ways. One to see a straight comparison of clockspeed and GB of RAM and another test to see how the new base model "Harpertown" 8-core compares to the previous "Clovertown" 8-core.

I did find that the machine tested would have cost $3,997 and not include Bluetooth, a 320GB HDD and an extra 1GB RAM. Anyone who bought a Mac Pro before Tuesday and within the last two weeks should return it.

right they are not G5s, it's that I have a G5 and it looks the same so my brain screwed up.

[digitalclips]

Originally Posted by jragosta

No. That would be insanely stupid.

(BTW, Apple hasn't sold G5s for quite a while. I'm assuming you meant Intel chips.)

Yep Intel ... My bad

But I am the one who bought a Lisa, an Apple ///, a Mac IIFX, and a Quadra 840 av, so just as I reached for the phone to order the 8 core I hesitated ...

[donebylee]

Originally Posted by solipsism

This makes this new machine a future-proof investment.

No such thing.

But I have ordered one because it should offer a good platform to carry me at least 3-5 years into the future.

I expect that at this time next year we will be looking at new Mac Pros with Nehalem, and they will look awfully good. But again the change will be evolutionary not revolutionary. Just like this one.

[digitalclips]

Originally Posted by donebylee

No such thing.

But I have ordered one because it should offer a good platform to carry me at least 3-5 years into the future.

I expect that at this time next year we will be looking at new Mac Pros with Nehalem, and they will look awfully good. But again the change will be evolutionary not revolutionary. Just like this one.

Stop tempting me ....

[melgross]

Originally Posted by donebylee

No such thing.

But I have ordered one because it should offer a good platform to carry me at least 3-5 years into the future.

I expect that at this time next year we will be looking at new Mac Pros with Nehalem, and they will look awfully good. But again the change will be evolutionary not revolutionary. Just like this one.

More so than this one.

This will be a serious architectural change. Moving to Penyrn is a noticeable, but fairly minor improvement.

[donebylee]

Originally Posted by digitalclips

Stop tempting me ....

LOL, the worst part of it is I wanted the 8800 so I have to wait until February for the system I ordered to ship. Now that is the definition of hell in my book.

Originally Posted by melgross

More so than this one.

This will be a serious architectural change. Moving to Penyrn is a noticeable, but fairly minor improvement.

From an architecture point of view you are right. Whether or not it will be a performance revolution is yet to be seen, and I am more affected by performance than architecture.

[netdog]

Not only did they test without SSE4 (as OS X is waiting for 10.5.2 for that), but they don't plan on implementing SSE4 into their benchmarking even though Intel claims that the gains are quite dramatic. Think that we will have to look elsewhere for our MP2 benchmarks perhaps.

[Nicnac]

Originally Posted by solipsism

The new one still looks great and I find it hard to conceive of how they could improve on this timeless design. Now, if the heat and innards are reduced then perhaps it would warrant some changes to fully optimize the space, but I don't think it will. Just my opinion.

Before the iPhone came out, I thought my RAZR was cool looking. Before the 'where's the computer' iMacs came out, I thought the flat-panel iMac was the coolest ever. The current 'G5' towers are simply TOO big. Working Mac IT, delivering, and setting up these machines, I am honestly stressed that I am one day either going to break one when I fail to lift it, or break myself when I go down instead of the machine going up.

Quite simply, there is no reason for the case to be quite so big now that they are more efficiently cooled Intels. And, Apple could really stun us with a sleek, small tower and make the current suddenly look oh so old.

[melgross]

Originally Posted by donebylee

From an architecture point of view you are right. Whether or not it will be a performance revolution is yet to be seen, and I am more affected by performance than architecture.

Performancewise, it will be a much bigger step than this one.

I also expect Apple do do more with the mobo and the rest of the case.

[donebylee]

Originally Posted by melgross

Performancewise, it will be a much bigger step than this one.

I also expect Apple do do more with the mobo and the rest of the case.

You may well be right. I expect to see new Mac Pros at about the same time next year with the new chips and MBs. We will all know for sure then.

In the meantime, I am looking forward to my Mac Pro. February can't come fast enough. Of course, I wouldn't mind if Apple shipped it earlier than that. Wouldn't mind a bit.

BTW, for those that haven't seen it, Bare Feats has a buyer's advice article up over here with their take on the new Mac Pro and which video card you should get.

[melgross]

Originally Posted by donebylee

You may well be right. I expect to see new Mac Pros at about the same time next year with the new chips and MBs. We will all know for sure then.

In the meantime, I am looking forward to my Mac Pro. February can't come fast enough. Of course, I wouldn't mind if Apple shipped it earlier than that. Wouldn't mind a bit.

BTW, for those that haven't seen it, Bare Feats has a buyer's advice article up over here with their take on the new Mac Pro and which video card you should get.

We may see these new machines at, or shortly after, the ADC this year. These chips are scheduled to arrive in the third quarter. That could very easily mean that Apple will have them then.

If it will be this time next year, that would be different. But, the evidence is that we may see 32nm around then. So the cycle begins again.

[donebylee]

Originally Posted by melgross

We may see these new machines at, or shortly after, the ADC this year. These chips are scheduled to arrive in the third quarter. That could very easily mean that Apple will have them then.

If it will be this time next year, that would be different. But, the evidence is that we may see 32nm around then. So the cycle begins again.

That sounds suspiciously like what we were hearing about Penryn all through 2007.

Given Apple's recent track record with the Mac Pro, I would be surprised if we see a newer version before the year is out.

But we will see.

[melgross]

Originally Posted by donebylee

That sounds suspiciously like what we were hearing about Penryn all through 2007.

Given Apple's recent track record with the Mac Pro, I would be surprised if we see a newer version before the year is out.

But we will see.

No, it's not. Penyrn had never been thought of as a major new design. It's an improvement over the 65nm designs because of some modificationss, but nothing truly major.

It's Intel's now well known "Tick Tock" strategy. The die shrink, with some improvements, is the "Tick". The new architecture is the "Tock".

Nehalem is different from Penyrn. It's the "Tock".

[solipsism]

Originally Posted by melgross

No, it's not. Penyrn had never been thought of as a major new design. It's an improvement over the 65nm designs because of some modificationss, but nothing truly major.

Nehalem is different.

Architecturally I agree, but the inclusion of SSE4 makes for a very compelling reason to upgrade an older Mac Pro or PowerMac (assuming that Apple adds Penryn support soon).

[donebylee]

Originally Posted by melgross

No, it's not. Penyrn had never been thought of as a major new design. It's an improvement over the 65nm designs because of some modificationss, but nothing truly major.

It's Intel's now well known "Tick Tock" strategy. The die shrink, with some improvements, is the "Tick". The new architecture is the "Tock".

Nehalem is different from Penyrn. It's the "Tock".

No you missed my point, which is that I don't think we will see Nehalems in a mac Pro until this time next year. That is what I was commenting on being similar to the Penryn situation. Everyone was expecting them much sooner than they arrived. I think we will see the same thing with Nehalem.

[melgross]

Originally Posted by solipsism

Architecturally I agree, but the inclusion of SSE4 makes for a very compelling reason to upgrade an older Mac Pro or PowerMac (assuming that Apple adds Penryn support soon).

Only for a few programs. It will also take months for most software to be able to use the new instructions.

If Nehalem machines do appear for the ADC, then that would coincide with their ability to use SSE4 as well. Right now, there is only one or two programs that can use them, and they aren't optimized yet either.

Otherwise, speed improvements range from a couple percent to 10%. Only a very few areas are better than that. Penyrn's main advantage is with the portable models right now, because of the additional battery life as well as the moderate performance increase.

I'm not saying its terrible for the Mac Pro, just not close to what it will be with Nehalem, along with the other improvements we'll see..

[melgross]

Originally Posted by donebylee

No you missed my point, which is that I don't think we will see Nehalems in a mac Pro until this time next year. That is what I was commenting on being similar to the Penryn situation. Everyone was expecting them much sooner than they arrived. I think we will see the same thing with Nehalem.

I didn't miss your point. I addressed it. Nehalem is expected, according to Intel, late this year. It's actually expected in the third quarter.

Re-read post number 34.

I see no reason why Apple would hold out until next year.