Anyone else disappointed by the 2.26ghz 8 core Nehalem Mac Pro's performance?

MovieCutter said:

You're talking smack about Pros who buy a 2.8Ghz Octo Mac and all you could afford is the 2.26? For all your talk, I'd expect you'd pick up the 2.93 like the rest of us real Pros...have fun with that. I picked up a 2.8 Octo Mac, 8GB RAM, 300GB 15k drive+750GB, 8800GPU, and 23" display for $1900 today...bad call? Not in the slightest. That being said, I also have a 2.93Ghz Octo on the way...thanks for your input about how buying a 2.8 Octo would be stupid though.

Click to expand...

I can afford whatever flavor of Mac Pro I wan't. I own 2 companies and we use all the latest high end Mac Pros and Dell Workstations. This 2.26 is just a computer to use for Vista, Office, VMARE, and some web design.

mogus3000 said:

Just took delivery of my 2.26ghz octo and i have been running some cinebench test as this machine is destined to become my cinema 4d (cinebench) workstation.

The test results are decent and it clearly outperforms the previous 2.8ghz and 3.0ghz in multithreading. Just a bit shy of the 3.2ghz. I have ran the the cinebench 4 times as it can vary quite a bit, but it always hovers above 20,000cb for multi-threading. I knew when i saw tessalators chart that the 2.26 - 18.000cb score was bull ****. The open gl score is not too hot though.

this machine is a tiny bit overpriced and the baseline octo should have been the 2.66 octo IMO, but i am very happy with it.
Remember folks, the previous gen 2.8Ghz octo represented insane value for money at the time. No other vendor could compete on price with the last model.

Click to expand...

The only CB benchmark I'd see so far for the 2.26 octo was more like 2300 on the single threaded so the above is more reassuring.

I ordered a 2.26 octo last Saturday and then saw the benchmarks and was getting concerned.

I do windows development for work but at home am planning to switch to a mac after buying a macbook 2.0 ghz. The macbook 2.0 ghz is more responsive than my home pc which is currently vista but a 3.2 core 2 duo.

I may use imovie and dabble in some video work for fun only so I'm thinking the 2.26 octo is sufficient. Will also be using ilife etc, browsing, occasional basic gaming - but most of my gaming is xbox or ps3 so no heavy gaming.

I also am going to learn Mac development so will use xcode etc.

I'm looking for responsiveness - like when you open apps, windows etc. - what I don't get on windows. Since I'm not going to do professional work on it i'm thinking this is the best investment since over time and with snow leopard, multi-threaded apps are the way to go.

For those with the 2.26 octo, how does it feel in general use?

Yes I am a little disappointed with the new mac pros and their prices, it would be nice to have the 2.66 at the 2.26 price point. The fact is that it isn't so I have decided to go for the new 2.26 over the 2.8. Sure I'm paying $200 more (the difference after I factor in upgrading ram, hd, ect in the 2008 model to equal the 2009 I'm getting) for a slightly "inferior" machine (according to current benchmarks) but I'm willing to make that sacrifice for newer technology. I do really hope with real world testing we see some better bench marks but I'm not counting on it. I'm happy with my decision as I hope anybody that has purchased a mac pro (whatever year) is.

superpalmtree said:

I can afford whatever flavor of Mac Pro I wan't. I own 2 companies and we use all the latest high end Mac Pros and Dell Workstations. This 2.26 is just a computer to use for Vista, Office, VMARE, and some web design.

Click to expand...

Check out http://www.gearspace.com/board/music-computers/371545-logic-pro-multicore-benchmarktest.html

This shows that for real time performance the old 2.8 octos running Logic Pro (judging by the comments on the thread) can wipe the floor with the new 2.26 octo.

Some people want a stable system that works with the software and hardware that they have now, not waiting months for the updated Logic or Pro Tools, or waiting for a new OS to make use of the (albeit great) feature that the new hardware has, and then waiting again for all of there software to be fully compatible.

For examle, if you wanted a new Pro Tools HD rig, you'd want hardware that was certified by Digidesign, and these new Mac Pros are not as yet certified, so can you not see that there would be wisdom in getting what you know is supposed to work.

Judging by the thread I've linked to up there I guess I am disappointed by the performance.

VirtualRain said:

It's hard to imagine that Ableton live would be highly multi-threaded. I would research how threaded this app is first. If it can only utilize a couple of cores or a few threads, then why get an Octocore? A faster 2.93 Quad would be more beneficial.

Click to expand...

From what I've read, Live 7 uses 8 cores and Live 8 will be able to use up to 32.

myca said:

Check out http://www.gearspace.com/board/music-computers/371545-logic-pro-multicore-benchmarktest.html

This clearly shows that for real time performance the old 2.8 octos running Logic Pro can (judging by the comments on the thread) can wipe the floor with the new 2.26 octo.

Click to expand...

I'm not looking to start an argument, but that is based on one user's experience.

Even the Cinebench scores have been different on the new Mac Pros. The chart made popular by Tesselator has the scores for the 2.26 at 3142 (single) and 20,138 (multiple) -- reported a few posts above.

Many other folks on the interwebz are reporting 2009 2x2.26 Nehalem scores over 3100 (single) and over 20,000 (multi) -- a far cry from the initial 2309 and 18,000 reported in the first day. Obviously not as high as the 2.93, but pretty dang fast.

'Just sayin, we need more data, but it does look like the often maligned 2.26 is faster than we first heard.

rbro said:

From what I've read, Live 7 uses 8 cores and Live 8 will be able to use up to 32.

Click to expand...

Live 7 did tout multiple core support as a feature, so if it's anything like Logic it should be able to use all cores in an efficient manner (like all DAWs should do).

You may want to look at what plug ins are being used though, some folk like NI are dragging their feet over multi core usage.

VirtualRain said:

The root of the problem is that Intel is charging a significant premium for Nehalem Gainstown processors, primarily because there is no competition from AMD and there's also only a very niche market for these processors.

Click to expand...

Let's get our facts straight. The processors that Apple is using the new Mac Pros are some of the cheapest XEON processors available. This has nothing to do with Intel's pricing and everything to do with Apple's decision to use the bottom-of-the-barrel XEON in the $2499 model and a mid-range XEON in the $3299.

There's no one to blame about the lack of bump in performance for the base model but Apple. They are making a much higher margin on both of these model because they picked cheaper chips this refresh.

New models (Early 2009)

Mac Pro "Quad-core", $2499, One 2.66Ghz Intel Xeon W3520 ($284) (Cheapest Nehalem XEON Intel offers) or 11 percent of total price.
Mac Pro "8 Core", $3299, Two 2.26Ghz Intel Xeon E5520's ($373 each, $746 total) (Mid-range XEON Intel offers in the "E-series" line) or 23 percent of total price.

Previous models (Early 2008)

Mac Pro "Two 2.8Ghz Quad-core", $2799, Two 2.8Ghz Intel XEON 5462 ($797 each, $1594 total) or 57 percent of total price.
Mac Pro "Two 3.0Ghz Quad-core", $3599, Two 3.0Ghz Intel XEON 5472 ($1022 each, $2044 total) or 57 percent of total price.

Apple's Nehalem XEON Choices

Dual Processor
Xeon w5580 (3.2 GHz) - > $1600;
Xeon x5570 (2.93 GHz) - > $1386;
Xeon x5560 (2.8 GHz) - > $1172;
Xeon x5550 (2.66 GHz) - > $958;
Xeon 5540 (2.53 GHz) - > $744;
Xeon e5530 (2.4 GHz) - > $530;
Xeon e5520 (2.26 GHz) - > $373;
Xeon e5506 (2.13 GHz) - > $266;
Xeon e5504 (2.0 GHz) - > $224;
Xeon e5502 - > $188.

Single Processor
Xeon w3570 (3.2 GHz) - > $999;
Xeon 3540 (2.93 GHz) - > $562;
Xeon w3520 (2.66 GHz) - > $284.

Grimace said:

I'm not looking to start an argument, but that is based on one user's experience.

Even the Cinebench scores have been different on the new Mac Pros. The chart made popular by Tesselator has the scores for the 2.26 at 3142 (single) and 20,138 (multiple) -- reported here: https://forums.macrumors.com/threads/665263/

Many other folks on the interwebz are reporting 2009 2x2.26 Nehalem scores over 3100 (single) and over 20,000 (multi) -- a far cry from the initial 2309 and 18,000 reported in the first day. Obviously not as high as the 2.93, but pretty dang fast.

'Just sayin, we need more data, but it does look like the often maligned 2.26 is faster than we first heard.

Click to expand...

Point taken, but users experiences can be handy to know, and I'm eager to hear more.

When either the quad G5s, or the octo Xeons came out Logic could only utilise half of the available cores, it was fixed with an update, and I guess early adopters can pay a heavy price sometimes. But that's why I was arguing that what is known to work in the here and now can be beneficial to some users.

I did find it odd that there were no comparisons of the new vs old machines on the Apple website useing Logic as a benchmark, something that they have done pretty much since they bought out Emagic. That as a user of Logic Pro did trouble me as I automatically thought that either the increase of performance is negligible, or could be worse. Either way if Apple ever decide to update Logic it may see massive gains with the new machines, but for now for some people the older model would be a safer bet.

Wow...this is a funny thread. I am new here and certainly did not expect to see this. Its funny because some people here are calling one another names over minute differences in performance. And its also funny to see some refer to the previous computer as "Old technology" when just last week it was new technology and will still smoke most of the competition. Oh..wait..Mac Pro's don't really have any competition. If you think that way, then the Nehalem's will soon be old technology because its safe to assume that Intel already has its replacement planned. I recently purchased a 2x2.8 Pro in January for my photography business, so I guess I fall into the "stupid enough" class. I am in no way sorry I did. The new 2x2.26 looks interesting, but the fact is that it really looks to be a lateral move. If I were buying today, I would probably look to pick up a 2x2.8 because you can probably get some deals. If there were no deals to be had, then I would pick up a new 2x2.26. Early benchmarks show no real gain, or anything that anybody would really feel in actual performance. So many people get hung up on benchmarks and so many of the differences can not even be felt in real world performance.

jaw04005 said:

Let's get our facts straight. The processors that Apple is using the new Mac Pros are some of the cheapest XEON processors available. This has nothing to do with Intel's pricing and everything to do with Apple's decision to use the bottom-of-the-barrel XEON in the $2499 model and a mid-range XEON in the $3299.

There's no one to blame about the lack of bump in performance for the base model but Apple. They are making a much higher margin on both of these model because they picked cheaper chips this refresh.

New models (Early 2009)

Mac Pro "Quad-core", $2499, One 2.66Ghz Intel Xeon W3520 ($284) (Cheapest Nehalem XEON Intel offers) or 11 percent of total price.
Mac Pro "8 Core", $3299, Two 2.26Ghz Intel Xeon E5520's ($373 each, $746 total) (Mid-range XEON Intel offers in the "E-series" line) or 23 percent of total price.

Previous models (Early 2008)

Mac Pro "Two 2.8Ghz Quad-core", $2799, Two 2.8Ghz Intel XEON 5462 ($797 each, $1594 total) or 57 percent of total price.
Mac Pro "Two 3.0Ghz Quad-core", $3599, Two 3.0Ghz Intel XEON 5472 ($1022 each, $2044 total) or 57 percent of total price.

Apple's Nehalem XEON Choices

Dual Processor
Xeon w5580 (3.2 GHz) - > $1600;
Xeon x5570 (2.93 GHz) - > $1386;
Xeon x5560 (2.8 GHz) - > $1172;
Xeon x5550 (2.66 GHz) - > $958;
Xeon 5540 (2.53 GHz) - > $744;
Xeon e5530 (2.4 GHz) - > $530;
Xeon e5520 (2.26 GHz) - > $373;
Xeon e5506 (2.13 GHz) - > $266;
Xeon e5504 (2.0 GHz) - > $224;
Xeon e5502 - > $188.

Single Processor
Xeon w3570 (3.2 GHz) - > $999;
Xeon 3540 (2.93 GHz) - > $562;
Xeon w3520 (2.66 GHz) - > $284.

Click to expand...

Great post, I wish this had been written up days ago when all the bickering was going on.

That, and the fact that the 2.26 has been inadvertently underrated already. Multi-core Cinebench performance actually approaches the old 3.2Ghz machines (though single performance is still lower).

Grimace said:

Great post, I wish this had been written up days ago when all the bickering was going on.

That, and the fact that the 2.26 has been inadvertently underrated already. Multi-core Cinebench performance actually approaches the old 3.2Ghz machines (though single performance is still lower).

Click to expand...

If someone at Barefeats (when they get these machines) can show me that the 2.26 is up-to par with the older 3.2 I'll hapily eat abit of humble pie over my recent rants, I'll do it in private of course

Edit, Barefeats has some early data that may not be as negative as some have thought, don't know if it's pie worthy yet though.

2.26 octo benchmarks

Hi, all!

Here are my benchmark results. I'm not at all disappointed, but my previous machine was an original MacBook Pro, so there is no real comparison there.

Machine is 2.26 octo w 6 GB memory, and 4870. Boot drive is Intel SSD and user volume is 3 velociraptors in Raid 0 stripe.

Geekbench http://browse.geekbench.ca/geekbench2/view/116463

Cinebench R10 and digLloydTools images are attached. The digLloydTools was run while Activity Monitor, Mail, iTunes, etc. was running. The machine was still very responsive, even when running 16 threads at 100% cpu. Very nice! Kind of strange to see the CPU displays running 16 threads at 100%.

superpalmtree said:

Yikes! That scotch explains why you were stupid enough to buy an old outdated Mac Pro 2.8. You may want to make purchases when sober.

Click to expand...

Unless it's the old 4 core model for about 30% less than the new 4 core model, those old 8 core Xeon systems are going to be a much faster option than the 2.66Ghz 4 Core Xeons you get for about the same price now if you're using a lot of multi-threaded applications.

As a side note, my only drunken purchases are on iTunes.

Thankfully, no Nickelback or Kid Rock yet but I bought an episode of Family Guy on iTunes at about 2am on a Sunday morning after a night on the ale (well Lager, absinthe, tequila, a little vodka...) and even though it started to seem very familiar, it took till about half way through for it to dawn on me that I already have it on DVD and just hadn't seen it in ages!

IainH said:

So youre running FCP I assume? Didnt realise it could use so many cores during render.

What else are you using all that power for?

Click to expand...

I don't think so. But it still runs slow. I think they have to rewrite to take advantage of more RAM and extra processors.

myca said:

If someone at Barefeats (when they get these machines) can show me that the 2.26 is up-to par with the older 3.2 I'll hapily eat abit of humble pie over my recent rants, I'll do it in private of course

Edit, Barefeats has some early data that may not be as negative as some have thought, don't know if it's pie worthy yet though.

Click to expand...

The 2.26 has been unfairly underrated so far, but I don't think it will surpass the 3.2Ghz overall. The 2.26 norms seem to be:

3,100-3,200 for single-core CineBench
19,000-20,000 for multi-core CinBench

Keep your pie!

DiskTester benchmarks

I just ran the disktester benchmarks.

The "OS" volume is an Intel SSD. The "User" volume is 3 x velociraptor 300gb in a Raid 0 stripe.

Looking like I expected here.


bash-3.2$ disktester show-info
DiskTester 2.0 (C) 2003-2006 diglloyd, Inc. All Rights Reserved

3 volumes: "OS", "User", "DiskTester_Install"

KB=1024 bytes, MB=1024KB, GB=1024MB, TB=1024GB

------------------------------------- "OS" -------------------------------------
Volume name: "OS"
Create date: Tuesday, March 10, 2009 5:49:46 AM PT
Modify date: Wednesday, March 11, 2009 3:19:29 PM PT
Backup date: Thursday, December 31, 1903 4:00:00 PM PT
Checked date: Thursday, December 31, 1903 4:00:00 PM PT
File count: 507152
Folder count: 128108
Total bytes: 79682387968 (74.2GB)
Free bytes: 48641122304 (45.3GB, 61.0%)
Block size: 4K (4096 bytes)
Total blocks: 19453708
Free blocks: 11875274 (61.0%)
Next allocation: 0
Rsrc fork clump size: 4K (4096 bytes)
Data fork clump size: 4K (4096 bytes)
Next catalog ID: 0
Flags: 4020 = {default volume}
File system ID: HFS Plus
Signature: H+ (HFS Plus)
Drive number: 1
Driver ref num: 0

Largest contiguous block: 44.0GB

------------------------------------ "User" ------------------------------------
Volume name: "User"
Create date: Tuesday, March 10, 2009 3:40:13 PM PT
Modify date: Wednesday, March 11, 2009 3:19:22 PM PT
Backup date: Thursday, December 31, 1903 4:00:00 PM PT
Checked date: Thursday, December 31, 1903 4:00:00 PM PT
File count: 21602
Folder count: 2621
Total bytes: 899175186432 (837.4GB)
Free bytes: 801406861312 (746.4GB, 89.1%)
Block size: 4K (4096 bytes)
Total blocks: 219525192
Free blocks: 195655972 (89.1%)
Next allocation: 0
Rsrc fork clump size: 4K (4096 bytes)
Data fork clump size: 4K (4096 bytes)
Next catalog ID: 0
Flags: 4000 = {}
File system ID: HFS Plus
Signature: H+ (HFS Plus)
Drive number: 1
Driver ref num: 0

Largest contiguous block: 741.1GB

Command "show-info" executed in 3.08 seconds.

bash-3.2$ disktester run-area-test OS
DiskTester 2.0 (C) 2003-2006 diglloyd, Inc. All Rights Reserved

=> Allocating maximum size contiguous file on "OS" (74.2GB)...44.0GB (59.3% of volume size)

NOTE:
(1) Contiguous space equal to 59.3% of volume "OS" was allocated. Depending on the location of the contiguous space (which disktester cannot determine), TOP SPEED MIGHT BE AS MUCH AS 32.6% FASTER (worst case, sequential). To avoid this ambiguity, test a freshly-formatted volume.

=> Using test size of 512MB, 4MB at a time, across a 44.0GB test file.
=> Testing at: 0%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%

Area 0% (offset 0B): writing...79.7MB/sec, reading...265MB/sec
Area 10% (offset 4.40GB): writing...79.0MB/sec, reading...265MB/sec
Area 20% (offset 8.80GB): writing...81.0MB/sec, reading...264MB/sec
Area 30% (offset 13.2GB): writing...80.5MB/sec, reading...264MB/sec
Area 40% (offset 17.6GB): writing...80.7MB/sec, reading...265MB/sec
Area 50% (offset 22.0GB): writing...78.9MB/sec, reading...260MB/sec
Area 60% (offset 26.4GB): writing...80.7MB/sec, reading...265MB/sec
Area 70% (offset 30.8GB): writing...79.4MB/sec, reading...265MB/sec
Area 80% (offset 35.2GB): writing...79.4MB/sec, reading...261MB/sec
Area 90% (offset 39.6GB): writing...79.6MB/sec, reading...260MB/sec
Area 100% (offset 43.5GB): writing...79.5MB/sec, reading...265MB/sec

======== Wednesday, March 11, 2009 3:20:39 PM PT, volume "OS" (74.2GB) =======


------------------ Averages for "OS" (512MB/4MB, 1 iteration) ------------------
Area (44.0GB) Write MB/sec Read MB/sec
0% 79.7 265
10% 79.0 265
20% 81.0 264
30% 80.5 264
40% 80.7 265
50% 78.9 260
60% 80.7 265
70% 79.4 265
80% 79.4 261
90% 79.6 260
100% 79.5 265
Average write speed across the volume: 79.9MB/sec
Average read speed across the volume: 263MB/sec


Command "run-area-test" executed in 92.13 seconds.

bash-3.2$ disktester run-area-test User
DiskTester 2.0 (C) 2003-2006 diglloyd, Inc. All Rights Reserved

=> Allocating maximum size contiguous file on "User" (837.4GB)...741.1GB (88.5% of volume size)

NOTE:
(1) Contiguous space equal to 88.5% of volume "User" was allocated. Depending on the location of the contiguous space (which disktester cannot determine), TOP SPEED MIGHT BE AS MUCH AS 9.2% FASTER (worst case, sequential). To avoid this ambiguity, test a freshly-formatted volume.

=> Using test size of 512MB, 4MB at a time, across a 741.1GB test file.
=> Testing at: 0%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 100%

Area 0% (offset 0B): writing...331MB/sec, reading...354MB/sec
Area 10% (offset 74.1GB): writing...333MB/sec, reading...330MB/sec
Area 20% (offset 148.2GB): writing...325MB/sec, reading...317MB/sec
Area 30% (offset 222.3GB): writing...316MB/sec, reading...310MB/sec
Area 40% (offset 296.5GB): writing...304MB/sec, reading...308MB/sec
Area 50% (offset 370.6GB): writing...293MB/sec, reading...287MB/sec
Area 60% (offset 444.7GB): writing...285MB/sec, reading...281MB/sec
Area 70% (offset 518.8GB): writing...272MB/sec, reading...268MB/sec
Area 80% (offset 592.9GB): writing...248MB/sec, reading...249MB/sec
Area 90% (offset 667.0GB): writing...225MB/sec, reading...236MB/sec
Area 100% (offset 740.6GB): writing...351MB/sec, reading...340MB/sec

====== Wednesday, March 11, 2009 3:22:19 PM PT, volume "User" (837.4GB) ======


----------------- Averages for "User" (512MB/4MB, 1 iteration) -----------------
Area (741.1GB) Write MB/sec Read MB/sec
0% 331 354
10% 333 330
20% 325 317
30% 316 310
40% 304 308
50% 293 287
60% 285 281
70% 272 268
80% 248 249
90% 225 236
100% 351 340
Average write speed across the volume: 299MB/sec
Average read speed across the volume: 298MB/sec


Command "run-area-test" executed in 41.15 seconds.

bash-3.2$ disktester run-sequential-suite OS
DiskTester 2.0 (C) 2003-2006 diglloyd, Inc. All Rights Reserved

=> Allocating maximum size contiguous file on "OS" (74.2GB)...44.0GB (59.3% of volume size)

NOTE:
(1) Contiguous space equal to 59.3% of volume "OS" was allocated. Depending on the location of the contiguous space (which disktester cannot determine), TOP SPEED MIGHT BE AS MUCH AS 32.6% FASTER (worst case, sequential). To avoid this ambiguity, test a freshly-formatted volume.

=> Using test size of 512MB at start (0%), within a 44.0GB test file.


Chunk Size 256K, test area start: writing...72.8MB/sec, reading...204MB/sec
Chunk Size 512K, test area start: writing...76.9MB/sec, reading...228MB/sec
Chunk Size 1MB, test area start: writing...75.3MB/sec, reading...227MB/sec
Chunk Size 2MB, test area start: writing...77.5MB/sec, reading...254MB/sec
Chunk Size 4MB, test area start: writing...79.1MB/sec, reading...263MB/sec
Chunk Size 8MB, test area start: writing...80.9MB/sec, reading...269MB/sec
Chunk Size 16MB, test area start: writing...77.7MB/sec, reading...266MB/sec
Chunk Size 32MB, test area start: writing...79.7MB/sec, reading...270MB/sec
Chunk Size 64MB, test area start: writing...76.1MB/sec, reading...180MB/sec
Chunk Size 128MB, test area start: writing...77.5MB/sec, reading...269MB/sec
Chunk Size 256MB, test area start: writing...79.7MB/sec, reading...271MB/sec

======== Wednesday, March 11, 2009 3:24:30 PM PT, volume "OS" (74.2GB) =======

----------------- Averages for "OS" (512MB/start, 1 iteration) -----------------
Chunk Size Write MB/sec Read MB/sec
256K 72.8 204
512K 76.9 228
1MB 75.3 227
2MB 77.5 254
4MB 79.1 263
8MB 80.9 269
16MB 77.7 266
32MB 79.7 270
64MB 76.1 180
128MB 77.5 269
256MB 79.7 271


Command "run-sequential-suite" executed in 96.68 seconds.

bash-3.2$ disktester run-sequential-suite User
DiskTester 2.0 (C) 2003-2006 diglloyd, Inc. All Rights Reserved

=> Allocating maximum size contiguous file on "User" (837.4GB)...741.1GB (88.5% of volume size)

NOTE:
(1) Contiguous space equal to 88.5% of volume "User" was allocated. Depending on the location of the contiguous space (which disktester cannot determine), TOP SPEED MIGHT BE AS MUCH AS 9.2% FASTER (worst case, sequential). To avoid this ambiguity, test a freshly-formatted volume.

=> Using test size of 512MB at start (0%), within a 741.1GB test file.


Chunk Size 256K, test area start: writing...209MB/sec, reading...261MB/sec
Chunk Size 512K, test area start: writing...292MB/sec, reading...324MB/sec
Chunk Size 1MB, test area start: writing...331MB/sec, reading...332MB/sec
Chunk Size 2MB, test area start: writing...343MB/sec, reading...325MB/sec
Chunk Size 4MB, test area start: writing...348MB/sec, reading...331MB/sec
Chunk Size 8MB, test area start: writing...348MB/sec, reading...334MB/sec
Chunk Size 16MB, test area start: writing...349MB/sec, reading...338MB/sec
Chunk Size 32MB, test area start: writing...353MB/sec, reading...338MB/sec
Chunk Size 64MB, test area start: writing...357MB/sec, reading...345MB/sec
Chunk Size 128MB, test area start: writing...331MB/sec, reading...346MB/sec
Chunk Size 256MB, test area start: writing...353MB/sec, reading...343MB/sec

====== Wednesday, March 11, 2009 3:26:23 PM PT, volume "User" (837.4GB) ======

---------------- Averages for "User" (512MB/start, 1 iteration) ----------------
Chunk Size Write MB/sec Read MB/sec
256K 209 261
512K 292 324
1MB 331 332
2MB 343 325
4MB 348 331
8MB 348 334
16MB 349 338
32MB 353 338
64MB 357 345
128MB 331 346
256MB 353 343


Command "run-sequential-suite" executed in 38.09 seconds.

bash-3.2$

Boneoh said:

Hi, all!

Here are my benchmark results. I'm not at all disappointed, but my previous machine was an original MacBook Pro, so there is no real comparison there.

Machine is 2.26 octo w 6 GB memory, and 4870. Boot drive is Intel SSD and user volume is 3 velociraptors in Raid 0 stripe.

Geekbench http://browse.geekbench.ca/geekbench2/view/116463

Cinebench R10 and digLloydTools images are attached. The digLloydTools was run while Activity Monitor, Mail, iTunes, etc. was running. The machine was still very responsive, even when running 16 threads at 100% cpu. Very nice! Kind of strange to see the CPU displays running 16 threads at 100%.

Click to expand...

The HDD's and their configuration certainly don't hurt.

Grimace said:

The 2.26 has been unfairly underrated so far, but I don't think it will surpass the 3.2Ghz overall. The 2.26 norms seem to be:

3,100-3,200 for single-core CineBench
19,000-20,000 for multi-core CinBench

Keep your pie!

Click to expand...

Yep! That does seem to have been the case.

Thanks to you that's been corrected!

https://forums.macrumors.com/posts/7255566/

New Data for Nehalem and Penryns

Collected from around the interwebz. (The 2.26 got unfairly thrashed originally, but still won't outdo the 3.2 Penryn.)