AMD Phenom 9550 vs Intel Core 2 Quad Q6600 – 7-Zip, Cinebench, Handbrake, and RetroMark Results
Cinebench is a highly threaded benchmark utility based on the Cinema 4D rendering tool with CPU SIMD support. I used the R11.5 64-bit version for this test. You can get older versions R10, R11, R15, and R20 in a single zip file @ archive.org.
| Processor↓ | Phenom 9550 Stock (2.20 GHz) is: | Phenom 9550 OC (2.64 GHz) is: | C2Q Q6600 Stock (2.40 GHz) is: | C2Q Q6600 OC (3.00 GHz) is: |
| Phenom 9550 (Stock 2.20 GHz) compared to: | Par | Single: -15.27% Four: -15.34% Ratio: +0.24% | Single: -11.60% Four: -4.88% Ratio: +7.75% | Single: -29.07% Four: -24.47% Ratio: +6.92% |
| Phenom 9550 (OC 2.64 GHz) compared to: | Single: +9.97% Four: +18.18% Ratio: -0.24% | Par | Single: +3.32% Four: +12.40% Ratio: +7.50% | Single: -8.68% Four: -10.74% Ratio: +6.66% |
| C2Q Q6600 (Stock 2.40 GHz) compared to: | Single: +13.11% Four: +5.13% Ratio: -7.20% | Single: +4.16% Four: -11.03% Ratio: -6.97% | Par | Single: -19.76% Four: -20.60% Ratio: -0.77% |
| C2Q Q6600 (OC 3.00 GHz) compared to: | Single: +40.98% Four: +32.41% Ratio: -6.47% | Single: +19.44% Four: +20.40% Ratio: -6.25% | Single: +24.63% Four: +25.94% Ratio: +0.77% | Par |
Rendering with a single thread has the Q6600 with a solid 13% advantage stock for stock. It takes the overclocked Ph9550 to beat the stock Q6600 which is around 3%. But things take a notable turn once 4 threads are used for rendering. Like the Fritz Chess bench earlier, the Phenoms have superior scaling as you increase the threads. The stock Q6600’s lead is only 4% after you load the CPU’s with 4 threads. The Phenom scaled 4.13x while the C2Q was 3.87. Directly comparing the two has the Phenoms scaling up to 7.5% better than the Q6600. Showing again the superior quad-core implementation.
| Processor ↓ | Phenom 9550 @ 2.2 GHz is: | the C2Q Q6600 @ 2.2 GHz is: |
| The Phenom 9550 @ 2.20 GHz compared to: | Par (Single: 0.61) Par (Four: 2.53) Par (Ratio: 4.17) | Single: -3.17% Four: +3.68% Ratio: +7.75% |
| The C2Q Q6600 @ 2.20 GHz compared to: | Single: +3.27% Four: -3.55% Ratio: -7.20% | Par (Single: 0.63) Par (Four: 2.44) Par (Ratio: 3.87) |
Things are definitely closer on a clock for clock basis. The 6600 only has a 3% lead via single thread rendering. But once all the cores are loaded down, the Phenom is over 3.5% faster. Again with thread scaling favoring the Phenom.
Handbrake is a multi-threaded video transcoding tool and takes advantageo of SIMD CPU instructions. It allows you to convert media to various formats like H.264, H265, etc . I used the same .mp4 video file and converted to H.264 format. I then checked the log after the conversation was completed and grabbed the output for Average Frames Per Second each CPU was able to encode at. I used the 1.0.0 64-bit version.
| Processor↓ | Phenom 9550 Stock (2.20 GHz) is: | Phenom 9550 OC (2.64 GHz) is: | C2Q Q6600 Stock (2.40 GHz) is: | C2Q Q6600 OC (3.00 GHz) is: |
| Phenom 9550 (Stock 2.20 GHz) compared to: | Par | -16.30% | -20.42% | -36.76% |
| Phenom 9550 (OC 2.64 GHz) compared to: | +19.48% | Par | -4.91% | -24.42% |
| C2Q Q6600 (Stock 2.40 GHz) compared to: | +25.66% | +5.16% | Par | -20.53% |
| C2Q Q6600 (OC 3.00 GHz) compared to: | +58.14% | +32.34% | +25.84% | Par |
Another case that has the Q6600 far ahead. Stock for stock the Phenom is over 20% slower. Even the overclocked Phenom is almost 5% slower than the stock 6600. I did notice that although the Phenom supported instructions like SSE3, Handbrake didn’t use them while they did use them on the Core 2 Quad CPU’s. Not sure why they wouldn’t but it’s something I noticed and definitely could have a performance impact.
I wanted to have a clock for clock comparison for Handbrake but when the Windows installation crashed so did all the other files including the sample video file I used for the benchmarking. I tried to find and re-download it but I couldn’t quite find the exact one and therefore couldn’t include the clock for clock comparison. It’s obvious though that the Q6600 would have been much faster. Probably close to 20% faster than the Phenom clock for clock.
7-Zip is a very popular file compression and decompression program with a great multi-threading benchmark utility. The built-in benchmark tool provides compression and decompression results. I used the default dictionary size which is 32 MB.
| Processor↓ | Phenom 9550 Stock (2.20 GHz) is: | Phenom 9550 OC (2.64 GHz) is: | C2Q Q6600 Stock (2.40 GHz) is: | C2Q Q6600 OC (3.00 GHz) is: |
| Phenom 9550 (Stock 2.20 GHz) compared to: | Par | Compression: -15.30% Decompression: -14.60 | Compression: +5.28% Decompression: +10.22 | Compression: -13.02% Decompression: -10.60 |
| Phenom 9550 (OC 2.64 GHz) compared to: | Compression: +18.05% Decompression: +17.08% | Par | Compression: +24.30% Decompression: +29.06% | Compression: +2.67% Decompression: +4.68% |
| C2Q Q6600 (Stock 2.40 GHz) compared to: | Compression: -5.01% Decompression: -9.27% | Compression: -19.54% Decompression: -22.51% | Par | Compression: -17.40% Decompression: -18.88% |
| C2Q Q6600 (OC 3.00 GHz) compared to: | Compression: +14.97% Decompression: +11.85% | Compression: -2.60% Decompression: -4.47% | Compression: +21.05% Decompression: +23.28% | Par |
The Barcelona/Agena architecture is very strong for this task. Probably a combination of the L3 cache and on-die dual channel memory controller providing tons of low latency bandwidth is a big reason. Stock for stock the Phenom is 5% faster in compression and doubled that to 10% in decompression. Even with the overclocked Q6600 having a 13% higher clock speed, the overclocked Phenom was over 2.5% faster in Compression and almost 5% fast in decompression.
| Processor ↓ | Phenom 9550 @ 2.2 GHz is: | C2Q Q6600 @ 2.2 GHz is: |
| Phenom 9550 @ 2.20 GHz compared to: | Par (Compression: 9,067) Par (Decompression: 13,670) | Compression: +12.25% Decompression: +19.70% |
| C2Q Q6600 @ 2.20 GHz compared to: | Compression: -10.91% Decompression: -16.46% | Par (Compression: 8,077) Par (Decompression: 11,420) |
As expected, clock for clock the Ph9550 is superior. A very impressive 12% in compression and almost 20% in decompression. The Phenom CPU’s have had very few wins thus far but this is a pretty big one.
RetroMark Quick Bench is a simple little program I made myself specifically to very quickly get performance results of older computers like AMD K6-2’s, Pentiums, etc. It was designed to run on Windows 95, 98, ME, and Windows 2000. But it will run on really all Windows including Windows Vista, 7, and 10. Since the program uses API specific code for things like GDI, comparisons must be from the same OS, video card and drivers. But anyways I may spiffy up the interface with some other tweaks and release it to be downloaded. So I won’t give any real details on this page. Above is a screen shot of me messing around and running it on my AMD Ryzen 1600 “AF” overclocked to 4 GHz (before I recently upgraded to an Alder Lake Intel i5 12400 setup) in Windows 10 but the results are not at all comparable to the CPU’s in this review do to the operating system, video card, etc. being completely different.
I only ran this just because. Anyways, I only compared the total score or RetroMark score.
| Processor↓ | Phenom 9550 Stock (2.20 GHz) is: | Phenom 9550 OC (2.64 GHz) is: | C2Q Q6600 Stock (2.40 GHz) is: | C2Q Q6600 OC (3.00 GHz) is: |
| Phenom 9550 (Stock 2.20 GHz) compared to: | Par | -16.55% | -8.5% | -36.65% |
| Phenom 9550 (OC 2.64 GHz) compared to: | +19.83% | Par | +9.64% | -12.30% |
| C2Q Q6600 (Stock 2.40 GHz) compared to: | +9.30% | -8.80% | Par | -20.01% |
| C2Q Q6600 (OC 3.00 GHz) compared to: | +36.64% | +14.02% | +25.01% | Par |
Comparing stock for stock the Q6600 has a solid +9% lead and the overclocked Q6600 is +14% faster than the overclocked Phenom. The results echo many other results thus far.
| Processor ↓ | Phenom 9550 @ 2.2 GHz is: | C2Q Q6600 @ 2.2 GHz is: |
| Phenom 9550 @ 2.20 GHz compared to: | Par (1,533.07) | -0.35% |
| C2Q Q6600 @ 2.20 GHz compared to: | +0.35% | Par (1,538.52) |
Clock for clock there is little difference. A couple of the tests are graphics tests that use the GDI APIs which are GPU accelerated. Since both CPU’s are using the same video card and drivers, the cube and fractal benchmark shouldn’t show much of a difference.