Apple Gets Nehalem Early, h264 and VMWare Performance Boosts

[MacRumors]

The release of the new Mac Pro on Tuesday marked the first use of Intel's Nehalem processor in Apple's products. As in the past, Intel has allowed Apple to get early access to their newest processors ahead of the competition. These Nehalem Xeon processors used in the high end Mac Pros have not even been officially announced by Intel yet.

Apple details the technical improvements of these new Nehalem processors on their product pages for the Mac Pro. While the descriptions and benchmarks are primarily marketing materials, they do offer simple explanations of some of the new technologies found in Nehalem. Some highlights include:

- Single die 64-bit architecture with fast access to cache data
- Integrated memory controller with significantly more memory bandwidth
- Turbo Boost: "If you’re using an application that doesn’t need every core, Turbo Boost shuts off the idle cores while simultaneously increasing the speed of the active ones, up to 3.33GHz on a 2.93GHz Mac Pro."

While Apple's tests show large improvements in memory bandwidth and floating point performance, many customers are awaiting 3rd party benchmarks to make a final purchasing decision. Notably, however, specific tasks or applications could see significantly higher performance boosts with Nehalem than might otherwise be expected.

An x264 developer has reported that Nehalem SSE changes are extremely beneficial to x264 performance and "have led to an enormous overall performance increase" over Penryn processors. As this processor support trickles out, it should speed up the time to encode h264 video substantially.

Meanwhile, VMWare customers may also see significant improvements in running VMWare Fusion on the new Nehalem Mac Pros. According to a forum post by VMWare's Ben Gertzfield, VMWare 2.02 already supports a new feature called "Extended Page Tables" which should result in "a pretty significant performance boost on the new Nehalem CPUs when running Fusion virtual machines."

This is a huge benefit to virtualization software: without EPT, a big chunk of the heavy lifting that a virtual machine has to do is emulating the "map virtual memory address X to physical memory address Y" work that a traditional MMU does.

The first of the Nehalem Mac Pros are expected to ship early next week.

Article Link: Apple Gets Nehalem Early, h264 and VMWare Performance Boosts

 

[cmendill]

Turbo!

Has anyone confirmed if the new Mac Pros have a Turbo button?

 

[adamw]

Good to hear! Way to go Apple releasing the new Mac Pros!

 

[iMacmatician]

Good to know, encouraged at the performance boosts since clock speeds are generally lower.

 

[kresh]

Has anyone confirmed if the new Mac Pros have a Turbo button?

A similar image popped into my head when reading this article as well. hehe You would think they would have used something other than "Turbo". Welcome to the 80's

 

[Wiska]

Ah, those days of my 12Mhz 386 PC with the turbo button that increased speeds to 16Mhz!! Those were the days.

 

[commander.data]

I don't doubt 2 x 2.93Hz Nehalem will beat 2 x 3.2GHz Harpertown.

The question is will 1 x 2.66GHz Nehalem, with only 4 cores, beat 2 x 2.80GHz Harpertown with 8 cores? In multithreaded applications or multitasking that can actually use 8 cores I doubt it. Which is why Apple doesn't should that comparison in it's benchmarks. Nehalem is fast, and HT maybe worth 20% increase on average, but it's hard to see it overcoming a 2 times core count advantage. I'd love to see the third-party benchmarks one way or another.

 

[Rhalliwell1]

XServe?

When will we see Nehalem XServes? One would presume fairly soon.

 

[SactoGuy18]

Ah, those days of my 12Mhz 386 PC with the turbo button that increased speeds to 16Mhz!! Those were the days.

You meant a 286 system. Intel never produced a 12 MHz 386 CPU officially--the first 80386 CPU started at 16 MHz.

And now, my home computer's Intel Pentium Dual-Core E2200 runs at 2,200 MHz.

 

[talkingfuture]

Sounds like a regular speed machine.

 

[the vj]

When I hear Apple talking about "speed improvement" for me that is a an amount of crap the size of jupiter.

I see all those benchmark showing "3 time faster than the previous processor", when the reality is that is barely noticiable.

The only way for you to get those results is to buy a Mac 6 revisions later at least, or lets say, wait 3 years.

I had my quad G5 and the only time I saw a realistic speed improvement was when I got the last quad core 2.8... THEN I saw an speed improvement.

But if you think the new ones are faster than the one I just bough 3 months ago... forget about it.

It is a huge pile of marketing BS unless you have one of the first Intel machines, THEN is when NOW you will see an improvement.

The benchmarks are such waiste of time, nothings more unrealistic in this world.

 

[poxonyou]

Bravo. You managed to be at the forefront of the one the category most of your customers cannot afford, while leaving your consumer lines at about the same exact specs they were at this time last year.

Well, I'm sure the few Richie Rich's here may be excited, and that seems to be the only people Apple cares about these days.

 

[SimonJ]

Turbo boost

Wow, am I the only one who instantly thought of Knight Rider?

 

[diamond.g]

Wow, am I the only one who instantly thought of Knight Rider?

You are not.

 

[Evangelion]

Wow, am I the only one who instantly thought of Knight Rider?

No you aren't

 

[iJaz]

Bah, I really want to upgrade my PowerMac G5 but this new MacPro revision is really not appealing to me. Maybe next revision, or the one after that, or...
I am really looking forward to real world benchmarks, but I suspect I'll save my money for a while.

 

[rubberduck007]

Now this man (woman?) speaks a lot of sense... well done Sir (Ma'am)...

When I hear Apple talking about "speed improvement" for me that is a an amount of crap the size of jupiter.

I see all those benchmark showing "3 time faster than the previous processor", when the reality is that is barely noticiable.

The only way for you to get those results is to buy a Mac 6 revisions later at least, or lets say, wait 3 years.

I had my quad G5 and the only time I saw a realistic speed improvement was when I got the last quad core 2.8... THEN I saw an speed improvement.

But if you think the new ones are faster than the one I just bough 3 months ago... forget about it.

It is a huge pile of marketing BS unless you have one of the first Intel machines, THEN is when NOW you will see an improvement.

The benchmarks are such waiste of time, nothings more unrealistic in this world.

 

[Cleve]

Well that beats marginal gains by iMac and Mac Mini. Guess you'd expect something like this after waiting a year and a half.

 

[iMacmatician]

I don't doubt 2 x 2.93Hz Nehalem will beat 2 x 3.2GHz Harpertown.

The question is will 1 x 2.66GHz Nehalem, with only 4 cores, beat 2 x 2.80GHz Harpertown with 8 cores? In multithreaded applications or multitasking that can actually use 8 cores I doubt it. Which is why Apple doesn't should that comparison in it's benchmarks. Nehalem is fast, and HT maybe worth 20% increase on average, but it's hard to see it overcoming a 2 times core count advantage. I'd love to see the third-party benchmarks one way or another.

Even Apple's own benchmarks, which likely favor Gainestown, do not show ≥2x the performance of Harpertown in most cases (assuming the tasks are multithreaded/multitasking).

When I hear Apple talking about "speed improvement" for me that is a an amount of crap the size of jupiter.

Don't you mean Uranus?

 

[fiatlux]

When I hear Apple talking about "speed improvement" for me that is a an amount of crap the size of jupiter.

Some architectural changes make a bigger impact than others, and there are times when there are big jumps in performances between consecutive generations (think about the move from Pentium 4 to Core architecture in the PC world).

Apple publishes a lot more than synthetic benchmarks for their new Mac Pros here: http://www.apple.com/macpro/performance.html

Some application benchmarks are impressive and should translate into real-world benefits (for instance the 60 to 80% improvements in 3D or Math applications). On the other hand, the 20% increase reported for other apps are not worth an upgrade, I agree.

I am quite curious as to where will sit the new 4 core entry-level Mac Pro in terms of perfs.

 

[iMacmatician]

Bravo. You managed to be at the forefront of the one the category most of your customers cannot afford, while leaving your consumer lines at about the same exact specs they were at this time last year.

Don't forget the doubling of RAM and HD, which don't happen every update. But the CPU speed increases of the iMacs are quite dismal. Only half the line got a speed bump.

 

[fleshman03]

Has anyone confirmed if the new Mac Pros have a Turbo button?

What did those actually do? I was too young to actually remember using one of those for a period of time. Any why would someone choose to run a desktop is a slower, non-turbo mode?

Oh and give me a 2.66ghz Nehalem + 4/8gb Ram + 1tb HD in a MBP and then I will be happy. (Any chance that happening in 09?)

 

[diamond.g]

What did those actually do? I was too young to actually remember using one of those for a period of time. Any why would someone choose to run a desktop is a slower, non-turbo mode?

Oh and give me a 2.66ghz Nehalem + 4/8gb Ram + 1tb HD in a MBP and then I will be happy. (Any chance that happening in 09?)

If I remember correctly the turbo button runs the CPU at it's normal speed, and when deactivated runs the computer at a slower speed.

 

[Chaszmyr]

Other than accepting ECC RAM, has it been established how these extremely expensive Nehalem Xeon processors compare (in terms of both performance and features) to the much cheaper Nehalem Core i7 desktop processors? You know, the ones Apple should have put in the new iMacs and didn't.

 

[Eidorian]

Nothing terribly new since Nehalem has been out for some time just not on a Mac.