CPU Performance

So my big focus with my testing this time around was to see how the 2600 and 2700 compare to the 2600X and 2700X as well as to see how the lower base clocks on the 2700 with a higher core count will compare to the 2600. The boost clocks on the 2700 are still higher so single core results should still be good, but I’m wondering if there will be some situations where the 2600 is the better option. So to start my testing I went with an older benchmark, the X264 HD Benchmark 4. X264 is an open source encoding format and this test runs a standard encode, noting the FPS that it encodes at then averaged over 4 tests. So here right out of the box, I ran into a situation where the 2600 was a slightly better option, by 6 FPS. This is one of the tests where higher core counts are used, but pure clock speed is favored. Both the 2700 and 2600 were still a good 20+ FPS lower than the much higher clocks of the 2600X and 2700X but still above even the 1800X from the original Ryzen launch.

graph1

For my next set of tests, I wanted to be able to compare both multi-core and single core results so I went with our recently added CPUz Benchmark. Here the test is run twice, once with just one core active and another with all of the cores. Now here is where the lower boost clock of the 2600 shows where the boost clock of the 2700 is up just behind the 2600X. In the multi-core test, things look a little more like how you would expect. The 2600 is a touch below the original 1600X but above the 1500X and the 2700 is well above the 2600X but below the 1700X.

graph2

graph3

For the second set of tests that look at both single core and multi-core, I went with Cinebench that renders a photo block by block. In the multi-core test, the 2600 and 1600X are right with each other, above the 7700K from Intel and the 2700 is paired up with the 1700X  and above the new 8700K. In the single-core test, the 2600 is up above all of the original Ryzen CPUs but still below all of the Intel CPUs in the test. The 2700 isn’t much higher, that extra boost clock helped a little though putting it just below the i5-8400.

graph4

graph5

My next tests are both math related, wPrime calculates prime out to 34 million and 7-Zip is a compression benchmark. In wPrime the lower the score is the better as it is the total time it took to run the test. Here the 2700 with its 8 cores really pulled ahead, coming in right in between the 1700X and 1700, also beating the 8700K as well. The 2600 did well beating the older 7700K but it is the last of the fast times before that big jump from around 140s to 170s and above. Now in 7-Zip the results are almost spot on the same, the 2600 is paired up with the 1600X again, above the 7700K and the 2700 is paired up with the 1700X and above the 8700K.

graph6

graph7

So JetStream is actually a compilation benchmark. This is a browser-based test designed to look at Java and HTML5 performance. It runs just about every browser-based test you can imagine then putting them all together for a score. Then it runs the total run 3 times to get an average. In the end, we get a good look at what you can expect for performance in the ever popular browser games. As usual, the AMD CPUs as a whole are in the bottom half of the chart, this is an area that Intel CPUs have always done better. That said the 2700 did well here but the 2600 didn’t fare so hot.

graph8

So next I got into Passmark’s Performance Test 9 and ran the CPU focused benchmark. This runs a stack of synthetic CPU benchmarks then takes those results to create an overall score. Here the 2700 did extremely well, coming in right with the always fast 1800X. The 2600 was once again paired up with the 1600X.

graph9

I ran PCMark10 because I love that they test using real-world situations. This includes video calls, web browsing, gaming, word processing, excel, photo editing, etc. Basically, everything I use my computers for lol. I included the overall score as well as all three of the test section results. The 2700 is about halfway down, just below the 1800X and above the 6900K. No big surprises here, The 1600X edged out the 2600 just slightly, but they are once again almost exactly the same for performance.

graph10

Before getting into actual gaming performance I did want to take a look at a few of my preferred synthetics. Well, 3DMark would fall under that, Dolphin 5 is actually an emulator and a great look at gaming performance in that situation. So in Dolphin the lower the score the better as it is actually the number of seconds it took for the test to run. This is another test where Intel has dominated by a large margin. The 2700 came in at 468 seconds outperforming the 1800X by a good amount but just behind the 2600X with its higher boost clock speeds. The 2600 was again close to the 1600X but this time did outperform it by 14 seconds. Overall if you are hoping to get the best performance in this Wii emulator Intel is still going to be the best option, but the new Ryzen options are still better than the old ones. Now in 3DMark, I ran the Fire Strike benchmark focusing on the physics score that is CPU dependent. The 2700 and 2600 don’t have much a gap between the two in this test but once again the 1600X is again right there with the 2600. The higher boost clocks of the 2600X and 2700X put them both much higher up in the charts though.

graph11

graph12

Okay now that we have taken a look at some Synthetic benchmarks, now we need to check out actual game performance. To do that I ran through four different games, not a huge amount but enough to get a peek at the performance. I tested using the newer Ashes of the Singularity with its CPU focused benchmark, Ghost Recon: Wildlands, Deus Ex: Mankind Divided, and the much much older TF2 to get a variety. My results were mixed. For starters in three of the tests, Intel is still dominating the top half of the charts but especially with the 2700X, AMD is clawing their way up slowly. Then in two of the four tests, the 2600 outperformed the 2700. Again these are specific situations where multiple cores are being used so the higher boost clock of the 2700 doesn’t come into play and the 2600 having a higher base clock shows. This was in Wildlands, where the 2600 was right up near the top behind the 2700X and the two new Intel CPUs and in Dues Ex. In TF2 the higher boost clock helped the 2700 get 9 more FPS and in Ashes of the Singularity, the 2700 had 3.4 FPS more than the 2600.

graph13

graph14graph15

graph16

For my last set of tests, I ran through a few extra benchmarks in AIDA64 focusing on Cache and memory speed and latency. We know the 2700 has 20MB of cache to the 19MB in the 2600. That difference was noticeable, especially in the L1 and L2 tests. The memory speeds and especially latency that the 2700X and 2600X both had big improvements on carried on to the 2600 and 2700. As for Single and Double precision FLOPS the 2700 with its two extra cores did come in above the 1600X, same with IOPS as well.

aida1

aida2

aida3

 

Log in to comment

VaporX's Avatar
VaporX replied the topic: #38527 01 Jun 2018 22:19
For gaming oriented builds I think the Ryzen 5 is a superior choice to the Ryzen 7. The performance delats are very close and the extra cores have minimal impact in a gaming system. The lower cost means more budget for graphics card, monitor or peripherals.

I am curious about your actual clock speeds. As I understand it the new new boost technology is more about the cooling solution than the work load and works to boost as many cores as it can within the cooling potential. I am seeing people say they are seeing around 3.8 Ghz on all cores under load with decent cooling, all the time.
garfi3ld's Avatar
garfi3ld replied the topic: #38528 02 Jun 2018 07:22
With the 2600, I'm seeing 4100MHz on single threaded applications and 3400 when using every core. The issue here is without manual overclocking, AMD still has to work within the limitations of the 65-watt TDP on both CPUs. So the new Precision Boost 2 works well, but is going to give you a lot more with the higher TDPs of the 2600X and 2700X.

HardOCP just posted up good overclocking coverage where they ran into the same TDP limitation as well. But the 2700 that they tested (and the 2600 that they haven't published) are both going to have similar overclocking room manually that the 2600X and 2700X have with PB2 and more, at a lower price even. Have to love all of the CPUs being unlocked :D

We have 1826 guests and no members online

supportus