Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
What settings? Without knowing your exact settings, you won't be able to benchmark it. You should skip 2-pass and just use the Apple TV prset—the best "bang for your buck" preset imo. I used to do 2-pass but it's really a waste of time for the little quality difference.

Too bad handbrake doesn't really scale beyond 4 cores—that's mainly a limitation of x264 (the underlying encoding library) so you won't see Handbrake taking advantage of Grand Central anytime soon, unless someone forks the project.

I'm using HB mainly for archival purposes so I'm looking for near lossless reproduction. I have a 1.5TB drive dedicated solely to media and in the near future I'll take the AppleTV or Mini plunge as a way to stream movies, pictures and music from my iTunes collection on my MP to my 52" Samsung LCD TV. I have a high-end amp, klipsch 5.1 surround system and everything is wired using HDMI. The idea is to use the MP as a media server and have an on-demand viewing experience for all my media.

Note, though, that x264 is full of asm optimizations for these new Nehalem chips which give a very significant speed boost at the same clockspeed compared to previous gen CPUs. Also, x264's issues with 8+ cores only come into play when using very low quality settings -- it's not that it can't take advantage of that many cores, just that the way it generates and closes out its threads has some overhead if the threads have very short lifetimes.

I could have sworn that HB was making use more than 4 of my cores. When ever it is running and I take a look at my activity monitor it always says in the CPU column between 400 and 500 and the number of threads is ~30. Maybe I'm just reading the monitor wrong.
 
In this example, the new 2.26GHz 8-core Nehalem performed comparably to previous 2.8GHz 8-core processor in multi-threaded tasks, but worse at single threaded tasks. Meanwhile, the multi-core performance of the new 2.93GHz processor significantly outpaced the previous generation machines.

The single-threaded results on the 2.26GHz machine don't make any sense. According to this chart, this machine has a speed-up of 7.83 going from single to eight core, while any other machine has only a speed up of a bit less than six (old machines) or a bit more than six (new machines). That's just not possible. I would doubt the single-threaded number. Maybe someone didn't watch out and Time Machine did a backup in the middle of the benchmark, or something similar stupid.
 
Can someone please try benchmarking Snow Leopard on these beasts - I'd love to know what THAT score will be!!

Nig.

Snow leopard does NOT run on Nehalem Mac Pro's yet. The latest dev seed supports only last generation Mac Pro's.
 
The single-threaded results on the 2.26GHz machine don't make any sense. According to this chart, this machine has a speed-up of 7.83 going from single to eight core, while any other machine has only a speed up of a bit less than six (old machines) or a bit more than six (new machines). That's just not possible. I would doubt the single-threaded number. Maybe someone didn't watch out and Time Machine did a backup in the middle of the benchmark, or something similar stupid.

That's nothing to be unexpected. Nehalem architecture's main strongpoint is utilizing multiple cores better than Penryn. That's why the new generation Mac Pro's speed up to almost 8x while the previous generation speeds up to 5.6-6x tops.

Don't expect single threaded performance of 2.26 to beat previous 2.8. It won't.
 
I could have sworn that HB was making use more than 4 of my cores. When ever it is running and I take a look at my activity monitor it always says in the CPU column between 400 and 500 and the number of threads is ~30. Maybe I'm just reading the monitor wrong.

1. On high-end machines, you are limited by the speed of the DVD drive. Especially if you use Handbrake with 2-Pass encoding, because it reads the DVD twice (good reason to use Mac The Ripper instead). You may have to rip a few DVDs with Mac The Ripper first and queue up the work to get to full CPU usage.

2. Handbrake may have many threads, but many don't do any of the hard work. If all the CPUs are used then the CPU column on an eight core machine should go to 800 percent. For example, you could have a thread that does nothing but to wait for the user to press some button, and that thread spends zero time. You can hundreds of that kind of thread running on a single core machine without problems.

That's nothing to be unexpected. Nehalem architecture's main strongpoint is utilizing multiple cores better than Penryn. That's why the new generation Mac Pro's speed up to almost 8x while the previous generation speeds up to 5.6-6x tops.

Don't expect single threaded performance of 2.26 to beat previous 2.8. It won't.

You didn't check the numbers.

2.26 = Speedup of 7.83.
2.93 = Speedup of 6 and a bit.

That's not possible. They use the same technology.
 
That's nothing to be unexpected. Nehalem architecture's main strongpoint is utilizing multiple cores better than Penryn. That's why the new generation Mac Pro's speed up to almost 8x while the previous generation speeds up to 5.6-6x tops.

Don't expect single threaded performance of 2.26 to beat previous 2.8. It won't.

I think his point though is that the 2.93GHz Nehalem doesn't perform as well (relatively). It's multi-core multiplier is not as high as the 2.26GHz. Which is a valid point.

We'll have to wait to see more numbers as more machines trickle in.

arn
 
Rating ? Positive or Negative

I've always wondered: who are the people who mark an article like this as 'Negative' on the main article page?

Nehalem Mac Pros Arrive: Unboxing and Benchmarks
...
Eg. Rating (38 Positives; 22 Negatives)

What is the rating supposed to measure ? What do we think of the new product ? What do we think of the writing style of the article ? How do we feel after purchasing a last gen mac pro a month ago ? My best guess is: how do we feel about something unrelated to the article. I can't see the connection.

Luke
 
1. On high-end machines, you are limited by the speed of the DVD drive. Especially if you use Handbrake with 2-Pass encoding, because it reads the DVD twice (good reason to use Mac The Ripper instead). You may have to rip a few DVDs with Mac The Ripper first and queue up the work to get to full CPU usage.

I realized this ages ago. And, although I haven't done this in a while, my recollection was that it took about the same time to encode directly from the drive as it did to rip first and then encode.

As a result I ALWAYS use MTR first so I have a digital file and then if I have multiple DVD's I can queue them up and let HB do it's stuff.

So, here's the question...if you want to force HB to use all 8 cores, can you duplicate HB and run 2 encodes in parallel? Or, will it keep trying to use the same 4 cores?
 
You didn't check the numbers.

2.26 = Speedup of 7.83.
2.93 = Speedup of 6 and a bit.

That's not possible. They use the same technology.

Ah, my bad. I thought you were only comparing to Penryn multipliers.

Yes there's something up there.
 
People have to make informed decisions about what they need to buy based on what they create with their computers.

I am a filmmaker and a DVD author, as well as a photographer. I am phasing out my old Dual 2.0 G5 and moving the Xeon Quad 3.0 to the second machine spot. For these reasons, I am upgrading to the top of the line. I will use the cores. My Quad Xeon was lurching earlier this year when I had to do days worth of HD encoding and crunching.

That being said, I also prefer to buy at the beginning of a machine's life cycle and not when it's a year old already. There are good deals out there - possibly - the older Octo machines, but I am not interested due to the kind of work I do and the imminent release of Snow Leopard.

On another, less quantifiable note, my main computer, in addition to its professional duties, is the most important machine in my life in many ways. It's the first thing I turn on in the morning and the last thing I turn off at night. I sit here all day working on it. I don't mind paying whatever I can afford to have it. I have always been able to sell my old machines for a decent price, as well.
 
this. was gonna buy the new one, but the dollar has gotten 50% more expensive so prices in my country are off the scale.

i paid 23k for my 8x2.8ghz now the 8x2.26 is 35k here. guess I was lucky to buy at the right time...

bummer, where do you live? sounds like high taxes also
 
Games are traditionally single-threaded (Half-Life 2 being just one famous example), because they require heavy synchronization and threading only adds to complexity, but not to performance. Games want ONE fast CPU core and a fast graphics card to be happy.

Is that not the same as two threads, then? (Not a hardware guy; honest question.)
 
These bench (they are cpu only) are impressive. lower cloks but more performance!!!

We will see another boost in 3D graphics benchs since the new videocards are much more powerfull than the older ones.
 
MacPro benchmark...

I can't wait till my 2.66 Quad MacPro comes in. Only 2 more days!!! I'm so glad i went with the Radeon HD4870. w00t w00t :D
 
The high bench marks of this Mac Pro sure got eclipsed quickly enough by a 32-core machine.

On other forums I've heard that you can easily build a do-it-yourself PC for a lot lot less money than this MacPro and get higher test bench results. Is this true and is it easily done? Apparently they're trying to point out that this Mac Pro is just an overpriced boat anchor. If the case is that I could build a more powerful computer for less money, could somebody direct me to a parts list to see how it's done? I know I couldn't get the Nehalem processors, because they're not available through retail yet.
 
I just noticed that all the clock speed fanatics who were ranting up and down about how the new systems were slower (they aren't, clearly), and that there was a massive price jump (there wasn't, in performance over performance), seem oddly silent today.
 
So this shows me that I will have to drop $6000 on a new machine to get top of the line/a machine that will give me noticeable improvement over the previous model.

Not a problem, I just don't have that kinda cash to fork over just yet, so I may be waiting another year with my dual core 2Ghz G5.
 
Benchmarks Question - Snow Leopard

Were those benchmarks done on Snow Leopard or just Leopard? From my understanding Snow Leopard suppose to improve multicore architecture which means that having 8 cores will give you better results that 4 cores. Plus, if more programs like Adobe Photoshop and Premier will jump on board of multi-core, in the end having more processors will benefit you.

Also, anyone knows if Mac Pro will be shipped with Snow Leopard? May be I miss understood, but the new ATI video card need Snow Leopard to run. Or maybe its just for old Mac Pros?
 
Not a problem, I just don't have that kinda cash to fork over just yet, so I may be waiting another year with my dual core 2Ghz G5.
$6000? No, I have a dual 2GHz G5, and any of these machines will be a noticeable improvement over your already fast machine.

If you're actually concerned about performance and not just having the fastest Mac available, though, you have no compelling reason to upgrade until after Snow Leopard comes out and the programs you use are rewritten for it.

Also, anyone knows if Mac Pro will be shipped with Snow Leopard? May be I miss understood, but the new ATI video card need Snow Leopard to run. Or maybe its just for old Mac Pros?
You misunderstood. Snow Leopard is not out yet. These machines are. How could they ship with it?

The ATI card you're thinking of needs 10.5.7 to run. Mac Pros ordered with those are not shipping yet, because 10.5.7 isn't finished yet (and, maybe, the cards aren't available yet).
 
I'm using HB mainly for archival purposes so I'm looking for near lossless reproduction. I have a 1.5TB drive dedicated solely to media and in the near future I'll take the AppleTV or Mini plunge as a way to stream movies, pictures and music from my iTunes collection on my MP to my 52" Samsung LCD TV. I have a high-end amp, klipsch 5.1 surround system and everything is wired using HDMI. The idea is to use the MP as a media server and have an on-demand viewing experience for all my media.



I could have sworn that HB was making use more than 4 of my cores. When ever it is running and I take a look at my activity monitor it always says in the CPU column between 400 and 500 and the number of threads is ~30. Maybe I'm just reading the monitor wrong.


It is worth it. I have both a MacMini and just got the Apple TV. Both are connected to my TV and I have my trusty G5 as a server. I cancelled DirectTV and have never looked back. The $90/month saving from DirectTv will quickly pay off my AppleTV. One think I would say is get the 160GB Apple TV. As you state with media the more space the better.
 
This benchmark proved that the lowered end new Mac Pro is a joke.


This post proved you didn't have a clue lol.

The single quad core it is still power as the old 8-core only in cpu power!!! These new machines shines with other apps thanks to nw gfxcards, memory speed and hyperthreading. And cost less. It is an impressive improvements, but also old 2,8Ghz are cool machines and even faster for single core apps.

I would buy the new quadcore over the old 8-core.
 
I understand that Snow Leopard is not out yet, but my new Mac Pro is schedualed to be shipped on March 12th. I also know about March 24th event which is suppose to be about software, which makes me think that they will probably release the Snow Leopard than. But if I get MP with ATI card that need Snow Leopard on March 13th, release on March 24th., makes me confused. Unless you dont need Snow Leopard to run ATI card or they will push shippment date.

But I got email from Apple saying that they are pushing shipping date from original 17th to 12th.
 
That did not last long. Already the new MP is no longer on the top of the benchmark list. IBM beat it with a score of 22,000. to Apple's 17,000

Also the Sun server that the mew Mac Pro beat was not Sun's top of the line. So I imagine Sun have one of their larger systems benchmarked. They might even beat IBM.

Still the Apple product looks good. Sun and IBM beat it using older Xeons but more of them and at much higher price.

To bad those old rumers about a Sun/Apple merger were wrong. Would be nice to be able to buy Mac OS X on Sun hardware. Sun covers the high end so much better than Apple and Apple has the desktop which Sun lacks
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.