PC Performance (January 2009)

It’s been over a year since I’ve updated the PC Performance chart. Why did it take so long? Well, over the last year I’ve made a lot of architectural changes, both to Geekbench and to the Geekbench Result Browser, that broke the chart generation scripts. After the Geekbench 2.1 release (and after the holidays) I had time to sit down and re-write the scripts, so here’s the latest installment of the PC Performance chart.

Setup

Results were collected from Geekbench 2 for Windows running on PCs with at least 512 MB of RAM and with processors that weren’t overclocked.

I’ve reported the average overall score for each processor. If you’re unfamiliar with Geekbench and how it measures performance, a score of 1000 is the score a Power Mac G5 @ 1.6GHz would receive. Higher scores are better.

Results

Intel Core i7 920
16 cores @ 2670 MHz		 7503
 
AMD Phenom II X4 940
4 cores @ 3000 MHz		 5875
 
Intel Core 2 Extreme X9770
4 cores @ 3200 MHz		 5271
 
Intel Core 2 Extreme X9650
4 cores @ 3000 MHz		 5107
 
Intel Core 2 Quad Q9550
4 cores @ 2830 MHz		 4925
 
Intel Core 2 Extreme Q6800
4 cores @ 2930 MHz		 4886
 
Intel Core 2 Quad Q9550
4 cores @ 2830 MHz		 4801
 
Intel Core 2 Extreme Q6850
4 cores @ 3000 MHz		 4791
 
AMD Phenom 9950
4 cores @ 2600 MHz		 4724
 
AMD Phenom 9850
4 cores @ 2500 MHz		 4412
 
Intel Core 2 Quad Q9400
4 cores @ 2660 MHz		 4298
 
Intel Core 2 Quad Q6700
4 cores @ 2660 MHz		 4259
 
Intel Core 2 Quad Q9300
4 cores @ 2500 MHz		 4185
 
AMD Phenom 9750
4 cores @ 2400 MHz		 4104
 
Intel Core 2 Quad Q8200
4 cores @ 2330 MHz		 4000
 
AMD Phenom 9650
4 cores @ 2300 MHz		 3965
 
Intel Core 2 Quad Q6600
4 cores @ 2400 MHz		 3960
 
AMD Phenom 9550
4 cores @ 2200 MHz		 3872
 
Intel Core 2 Duo E8600
2 cores @ 3330 MHz		 3855
 
AMD Phenom 9600
4 cores @ 2300 MHz		 3846
 
Intel Core 2 Duo E8600
2 cores @ 3330 MHz		 3794
 
AMD Phenom 9500
4 cores @ 2200 MHz		 3652
 
Intel Core 2 Duo E8500
2 cores @ 3160 MHz		 3451
 
AMD Phenom 8650
3 cores @ 2300 MHz		 3367
 
Intel Core 2 Duo E8400
2 cores @ 3000 MHz		 3343
 
Intel Core 2 Duo E6850
2 cores @ 3000 MHz		 3334
 
AMD Athlon 64 X2 6400+
2 cores @ 3200 MHz		 3307
 
Intel Core 2 Duo E8400
2 cores @ 3000 MHz		 3291
 
Intel Core 2 Extreme X6800
2 cores @ 2930 MHz		 3102
 
Intel Core 2 Extreme X7900
2 cores @ 2800 MHz		 3093
 
AMD Athlon 64 X2 6000+
2 cores @ 3000 MHz		 3083
 
Intel Core 2 Duo E7300
2 cores @ 2660 MHz		 3065
 
AMD Phenom 8450
3 cores @ 2100 MHz		 3063
 
Intel Core 2 Duo E6750
2 cores @ 2660 MHz		 3022
 
Intel Core 2 Extreme X9000
2 cores @ 2800 MHz		 2947
 
AMD Athlon 64 FX-62
2 cores @ 2800 MHz		 2923
 
Intel Core 2 Duo E6700
2 cores @ 2660 MHz		 2895
 
AMD Athlon 64 X2 5400+
2 cores @ 2800 MHz		 2894
 
Intel Core 2 Duo E7200
2 cores @ 2530 MHz		 2810
 
AMD Athlon 64 X2 5200+
2 cores @ 2600 MHz		 2691
 
Intel Core 2 Duo E6600
2 cores @ 2400 MHz		 2656
 
AMD Athlon 64 FX-60
2 cores @ 2600 MHz		 2650
 
Intel Core 2 Duo E6550
2 cores @ 2330 MHz		 2640
 
AMD Athlon 64 X2 5000+
2 cores @ 2600 MHz		 2612
 
AMD Athlon X2 4850e
2 cores @ 2500 MHz		 2603
 
AMD Athlon 64 X2 4600+
2 cores @ 2400 MHz		 2466
 
Intel Core 2 Duo E4700
2 cores @ 2600 MHz		 2440
 
Intel Pentium Dual E2200
2 cores @ 2200 MHz		 2418
 
AMD Athlon 64 X2 4800+
2 cores @ 2400 MHz		 2416
 
Intel Core 2 Duo E6420
2 cores @ 2130 MHz		 2412
 
Intel Core 2 Duo E6400
2 cores @ 2130 MHz		 2357
 
Intel Core 2 Duo E4400
2 cores @ 2000 MHz		 2206
 
Intel Pentium Dual E2180
2 cores @ 2000 MHz		 2200
 
Intel Core 2 Duo E4400
2 cores @ 2000 MHz		 2194
 
AMD Athlon 64 X2 4200+
2 cores @ 2200 MHz		 2183
 
AMD Athlon 64 X2 4400+
2 cores @ 2200 MHz		 2173
 
AMD Athlon X2 BE-2350
2 cores @ 2100 MHz		 2150
 
Intel Core 2 Duo E6320
2 cores @ 1860 MHz		 2111
 
AMD Athlon 64 X2 4000+
2 cores @ 2100 MHz		 2111
 
Intel Core 2 Duo E6300
2 cores @ 1860 MHz		 2102
 
Intel Core 2 Duo E4300
2 cores @ 1800 MHz		 2091
 
Intel Pentium D 945/950
2 cores @ 3400 MHz		 2065
 
AMD Athlon 64 X2 3800+
2 cores @ 2000 MHz		 2053
 
Intel Pentium Dual E2160
2 cores @ 1800 MHz		 2049
 
Intel Pentium D 840
2 cores @ 3200 MHz		 2006
 
AMD Athlon 64 X2 3600+
2 cores @ 2000 MHz		 1991
 
Intel Pentium D 935
2 cores @ 3200 MHz		 1938
 
Intel Pentium D 940
2 cores @ 3200 MHz		 1934
 
Intel Pentium 4 670/672
2 cores @ 3800 MHz		 1911
 
Intel Pentium D 830
2 cores @ 3000 MHz		 1886
 
Intel Pentium D 925
2 cores @ 3000 MHz		 1882
 
AMD Athlon 64 FX-55
1 core @ 2600 MHz		 1879
 
Intel Pentium D 925/930
2 cores @ 3000 MHz		 1843
 
AMD Athlon 64 3800+
1 core @ 2400 MHz		 1756
 
Intel Pentium 4 560
2 cores @ 3600 MHz		 1752
 
Intel Pentium D 820
2 cores @ 2800 MHz		 1741
 
Intel Pentium D 920
2 cores @ 2800 MHz		 1734
 
Intel Pentium 4 551
2 cores @ 3400 MHz		 1726
 
AMD Athlon 64 4000+
1 core @ 2400 MHz		 1711
 
Intel Pentium D 915
2 cores @ 2800 MHz		 1694
 
Intel Pentium D 805
2 cores @ 2660 MHz		 1685
 
AMD Athlon 64 3700+
1 core @ 2200 MHz		 1631
 
AMD Athlon 64 3500+
1 core @ 2200 MHz		 1584
 
AMD Athlon 64 3400+
1 core @ 2200 MHz		 1557
 
AMD Athlon 64 3200+
1 core @ 2000 MHz		 1446
 
Intel Pentium 4 630
2 cores @ 3000 MHz		 1443
 
Intel Pentium 4 541
2 cores @ 3200 MHz		 1406
 
Intel Pentium 4 520/620
2 cores @ 2800 MHz		 1375
 
AMD Athlon 64 3000+
1 core @ 1800 MHz		 1330
 
Intel Pentium 4 519K/524
1 core @ 3060 MHz		 1308
 
Intel Pentium 4 515/516
1 core @ 2930 MHz		 1256
 
AMD Athlon 64 2800+
1 core @ 1800 MHz		 1218
 
Intel Pentium 4 506
1 core @ 2660 MHz		 1044
 

Posted January 26th, 2009 in Geekbench, Hardware, Windows | Comments

Quad-Core Performance (December 2007)

Currently, AMD and Intel have different ideas on how to build a quad-core processor. Intel’s approach is to essentially stick two dual-core processors together on a single chip, while AMD’s approach is to build a true quad-core processor.

Both approaches have their advantages; Intel’s approach allowed them to bring quad-core processors to market much faster, while AMD’s approach should allow them to build higher performance quad-core processors.

Of course, the question is whether AMD’s “true” quad-core processors are faster than Intel’s quad-core processors. In order to answer that question I thought I’d take a look at Geekbench 2 results for two different systems; one built around an AMD Phenom 9600 and one built around an Intel Core 2 Quad 6600.

Setup

  • AMD Phenom

    • AMD Phenom 9600 @ 2.30 GHz
    • Gigabyte Technology Co., Ltd. GA-MA790FX-DS5
    • 4.00 GB 800 MHz
    • Microsoft Windows Vista Ultimate

  • Intel Core 2 Quad

    • Intel Core 2 Quad Q6600 @ 2.40 GHz
    • Gigabyte Technology Co., Ltd. 965P-S3
    • 4.00 GB 800 MHz
    • Microsoft Windows Vista Ultimate

I’m reporting the baseline score, rather than the raw score, for the benchmarks (where a score of 1000 is the score a Power Mac G5 1.6GHz would receive). Higher is better.

Results

Overall Performance

AMD Phenom 9600
 3851
 
Intel Core 2 Quad Q6600
 3813
 

Integer Performance

AMD Phenom 9600
 4402
 
Intel Core 2 Quad Q6600
 5120
 

Floating Point Performance

AMD Phenom 9600
 5038
 
Intel Core 2 Quad Q6600
 4039
 

Memory Performance

AMD Phenom 9600
 1951
 
Intel Core 2 Quad Q6600
 1943
 

Stream Performance

AMD Phenom 9600
 1574
 
Intel Core 2 Quad Q6600
 2191
 

Conclusions

Looking at the results, it’s not clear there’s an advantage to AMD’s “true” quad-core design, at least as implemented in the Phenom 9600. While the Phenom 9600 has a slightly lower clock frequency than the Q6600, and managed to outscore the Q6600 in a couple of benchmark sections, it was also outscored by the Q6600 in a couple of other benchmark sections, and received essentially the same score as the Q6600 overall.

When you consider that both processors are approximately the same price, it’s hard to recommend one over the other. Personally, I’d choose the Q6600 over the Phenom 9600 for no other reason than the Q6600 (and the accompanying motherboards) is much more mature. I’m not confident AMD has sorted out all of the initial problems with the Phenom.

Also, while the Phenom 9600 is the fastest quad-core desktop processor AMD offers, the Core 2 Quad Q6600 is the slowest quad-core desktop processor Intel offers. If you want the fastest quad-core processor available (and don’t mind spending a lot), you’ll want an Intel processor.

Posted December 15th, 2007 in AMD, Geekbench, Hardware, Intel | Comments

Sun Ultra 24

So, the Sun Ultra 24 I’m evaluating through Sun’s Try and Buy program arrived today. After hooking it up to the network it managed to knock every other host (including a Sun Ultra 20 M2) off the network. Next Solaris got stuck in an infinite loop of:

“Oh look, a network connection!”
“Oh dear, where’d my network connection go?”

Hoping it was a Solaris problem (hey, it’s not the first time I’ve had networking issues with Solaris on Sun hardware), I installed Windows Vista. And just like Solaris, Vista got stuck in an infinite loop of:

“Oh look, a network connection!”
“Oh dear, where’d my network connection go?”

I think it’s time to get in touch with Sun.

Posted December 4th, 2007 in Hardware | Comments

MacBook Pro Performance (November 2007)

Apple quietly updated the MacBook Pro along with the MacBook earlier this month.
The MacBook update was quite substantial; the updated added, among other things, the Santa Rosa chipset to the MacBook which improved performance dramatically. The MacBook Pro update wasn’t as substantial; it added a larger hard drive and a faster processor as build-to-order options.

What I (and others) have wondered is the faster processor (an Intel Core 2 Duo T7800 at 2.6GHz) worth the price? I’ve gathered Geekbench 2 results for all three MacBook Pro processor configurations to find out.

Setup

  • MacBook Pro (Mid 2007)
    • Intel Core 2 Duo @ 2.60GHz or
    • Intel Core 2 Duo @ 2.40GHz or
    • Intel Core 2 Duo @ 2.20GHz
    • 2.00 GB 667 MHz DDR2 SDRAM
    • Mac OS X 10.5.1 (Build 9B18)

I’m reporting the baseline score, rather than the raw score, for the benchmarks (where a score of 1000 is the score a Power Mac G5 1.6GHz would receive). Higher is better.

Results

Overall Performance

MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.60GHz		 3294
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.40GHz		 3094
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.20GHz		 2829
 

Integer Performance

MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.60GHz		 2865
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.40GHz		 2698
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.20GHz		 2446
 

Floating Point Performance

MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.60GHz		 4597
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.40GHz		 4292
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.20GHz		 3841
 

Memory Performance

MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.60GHz		 2435
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.40GHz		 2253
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.20GHz		 2186
 

Stream Performance

MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.60GHz		 1962
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.40GHz		 1974
 
MacBook Pro (Mid 2007)
Intel Core 2 Duo @ 2.20GHz		 1920
 

Conclusion

On the latest MacBook Pros, processor benchmark scores (Integer Performance and Floating Point Performance) scale with processor speed; a 20% increase in speed will bring about a 20% increase in score.

Memory benchmark scores (Memory Performance and Stream Performance) are a different story. Since all of the MacBooks have the same chipset and use the same memory (which influence memory benchmark scores far more than the processor), there isn’t as much of a gain.

So, is configuring a MacBook Pro with a 2.6GHz processor instead of a 2.4GHz processor worth it? If you’re running a lot of processor-intensive tasks where even a small increase in performance is noticeable (and appreciated) then you might want to consider it. Otherwise I’d recommend adding more RAM instead. More RAM will probably help your MacBook Pro performance more than just a faster processor.

Posted November 26th, 2007 in Geekbench, Hardware, MacBook Pro | Comments

MacBook Performance (November 2007)

Apple quietly released new MacBooks last week which feature (among other things) the Santa Rosa chipset and, for some models, a slightly faster processor. I thought it’d be interesting to compare two black MacBooks, the new model against the previous model, to see how performance has changed.

Setup

  • MacBook (Late 2007)

    • Intel Core 2 Duo @ 2.20GHz
    • 1.00 GB 667 MHz DDR2 SDRAM
    • Mac OS X 10.5 (Build 9A3110)

  • MacBook (Mid 2007)

    • Intel Core 2 Duo @ 2.00GHz
    • 1.00 GB 667 MHz DDR2 SDRAM
    • Mac OS X 10.5 (Build 9A581)

I’m reporting the baseline score, rather than the raw score, for each benchmark (where a score of 1000 is the score a Power Mac G5 1.6GHz would receive). Higher is better.

Results

Overall Performance

MacBook (Late 2007)
Intel Core 2 Duo @ 2.20GHz		 2908
 
MacBook (Mid 2007)
Intel Core 2 Duo @ 2.16GHz		 2760
 

Integer Performance

MacBook (Late 2007)
Intel Core 2 Duo @ 2.20GHz		 2580
 
MacBook (Mid 2007)
Intel Core 2 Duo @ 2.16GHz		 2511
 

Floating Point Performance

MacBook (Late 2007)
Intel Core 2 Duo @ 2.20GHz		 3988
 
MacBook (Mid 2007)
Intel Core 2 Duo @ 2.16GHz		 3890
 

Memory Performance

MacBook (Late 2007)
Intel Core 2 Duo @ 2.20GHz		 2153
 
MacBook (Mid 2007)
Intel Core 2 Duo @ 2.16GHz		 1878
 

Stream Performance

MacBook (Late 2007)
Intel Core 2 Duo @ 2.20GHz		 1789
 
MacBook (Mid 2007)
Intel Core 2 Duo @ 2.16GHz		 1449
 

Conclusions

While the new MacBooks have only modest processor performance gains over the previous MacBooks, the Santa Rosa chipset helps the new MacBooks achieve much more impressive performance gains over the previous MacBooks; memory performance is up almost 15% while stream performance (which relies heavily on memory) is up almost 25%.

When you consider the fact that the new MacBooks also support 4GB of RAM (the previous MacBooks only “officially” supported 2GB of RAM) the new MacBooks are a treat for anyone who runs memory-intensive applications but doesn’t want to pay the premium for a MacBook Pro.

Posted November 6th, 2007 in Geekbench, Hardware, MacBook | Comments

Matched RAM Performance

Last weekend I upgraded the RAM in my iMac from 1 GB (two 512 MB sticks) to 2 GB (two 1024 MB sticks) (in case you’re curious, I’m using Mushkin RAM). While it’s not been a huge improvement overall (Safari leaks memory just as fast as before), the extra RAM has made Aperture a lot faster; I no longer cringe when I start editing photos in Aperture.

While I was replacing the RAM I thought it might be interesting to see if it’s important to have matched RAM (in this case two 512 MB sticks or two 1024 MB sticks) or whether unmatched RAM (one 1024MB stick) runs just as fast. Of course, I’m using Geekbench 2 as my benchmark program.

Setup

  • iMac (Late 2006)
    • Intel Core 2 Duo @ 2.0GHz
    • 1.0 GB 667 MHz DDR2 SDRAM (2 x 512 MB) or
    • 1.0 GB 667 MHz DDR2 SDRAM (1 x 1024 MB) or
    • 2.0 GB 667 MHz DDR2 SDRAM (2 x 1024 MB)
    • Mac OS X 10.4.9 (Build 8P2137)

I’m reporting the baseline score, rather than the raw score, for each benchmark (where a score of 1000 is the score a Power Mac G5 1.6GHz would receive). Higher is better.

Results

Overall Performance

iMac (Late 2006)
2 x 512 MB		 2650
 
iMac (Late 2006)
1 x 1024 MB		 2607
 
iMac (Late 2006)
2 x 1024 MB		 2651
 

Integer Performance

iMac (Late 2006)
2 x 512 MB		 2243
 
iMac (Late 2006)
1 x 1024 MB		 2160
 
iMac (Late 2006)
2 x 1024 MB		 2244
 

Floating Point Performance

iMac (Late 2006)
2 x 512 MB		 3776
 
iMac (Late 2006)
1 x 1024 MB		 3794
 
iMac (Late 2006)
2 x 1024 MB		 3783
 

Memory Performance

iMac (Late 2006)
2 x 512 MB		 1939
 
iMac (Late 2006)
1 x 1024 MB		 1928
 
iMac (Late 2006)
2 x 1024 MB		 1936
 

Stream Performance

iMac (Late 2006)
2 x 512 MB		 1563
 
iMac (Late 2006)
1 x 1024 MB		 1383
 
iMac (Late 2006)
2 x 1024 MB		 1545
 

Conclusion

Overall, it doesn’t look like matched RAM offers a huge performance difference over unmatched RAM (save for the Stream benchmarks which measure raw memory bandwidth, in which case matched RAM is slightly better); if you have to choose between more RAM or matched RAM, I’d argue that you’re better off choosing more RAM (especially if you’re using Aperture).

Posted June 25th, 2007 in Geekbench, Hardware | Comments