We have all been waiting for this! AMD is now the best in absolutely everything! Right? Well, yes and maybe not exactly. But this is still a very exciting launch and there is a lot to digest. We will start with our coverage of the Ryzen 9 7950X 16 core chip that in my mind is an absolute productivity beast but also has some improvements to do elsewhere. Let’s get into it!
So 7950X is the highest end consumer CPU that AMD is currently launching. It features 16 cores and 32 threads with 4.5 GHz base clock and 5.7 GHz boost clock. This is a considerable jump from the last Gen 3.4 GHz base clock and 4.9 GHz boost clock.
AMD also mentions approximately 13% IPC uplift through improvements in branch prediction, load/store, front end as well as others.
Rather than digging deep into these, let’s focus on benchmarks! In order to do it properly, we have a few set-ups: the high-end chips from last gen from both Intel and AMD, as well as two new chips - the 8 core 7700X and 16 core 7950X.
First test is Blender where we load up all the cores to the max and take the CPU for a long and hard spin. In this test 7950X is almost 30% faster than the last gen 5950X and almost 31% faster than Intel 12900K.
When checking our frequency we find the Ryzen 9 7950X ramp just above 5 GHz and then slowly drop down to around 4740 MHz by the end of the test. Looking at CPU temperature we see it jumping straight to 95C and staying there. This is at TJMax temperature for Ryzen 7000 series and, according to AMD, is very much expected behaviour.
Quoting AMD here - “With the new AM5 socket and higher TDP, most processors will run into thermal wall before they hit power wall.”
To add to this, they also noted that “TJMAX is the max safe operating temperature rather than absolute max temperature. This processor is designed to run at TJMAX 24/7 without risk of damage or deterioration. At 95 degrees it is not running hot, rather it will intentionally go to this temperature as much as possible under load because the power management system knows that this is the ideal way to squeeze the most out of the chip without damaging it.”
A very interesting way of saying that it will run hot and you need plenty of cooling, but based on our results - cooling is needed for the chip to go above base clock and with exceptionally good cooling such as 360mm liquid AIO or equivalent, you are able to extract much more performance.
When we compare temperatures to the rest of the set-ups, we can see Intel is up top hugging the 100C line and both new Ryzen chips are at 95c, while last generation 5950X is hovering just above 70c.
If we look at the amount of power drawn, this is where it gets really interesting. New 7950X is sucking back over 220W at the start and then reducing down to around 180W towards the end of the render, while the Intel chip stays at around 230W for the whole duration and, as you may remember, completes the render 31% slower.
So the new chip is not only running at approximately 22% less power, that is after the initial ramp down - it also completed the work 31% faster. However when compared to 5950X - it is running with approximately 23% more power but still completes the work about 30% faster, so there is slight improvement here too.
With the power and thermals out of the way, let's now look at a few benchmarks to get the raw performance.
In the Blender benchmark covering three different renders - Monster, Junk Shop and Classroom we still have a massive lead in samples per minute rendered when compared to other chips.
In the Vray benchmark the new 7950X is just shy of 27% faster than its predecessor and 33.5% faster than its Intel counterpart.
For the 7zip benchmark we broke it out into Single Thread and Multi Threaded tests to check out performance of both. In Single core 7950X takes the lead by a few hundred points from 7700X as it clocks in above 5.74 GHz while 7700X maxed out at 5.56 GHz.
And in the Multi Threaded test there is yet another huge jump in performance from the last gen 5950X and almost double the performance on the decompression test to the rest of the chips. In compression the difference to Intel CPU is 26%.
I think it’s time to check out some games! Let’s start with good old Shadow of the Tomb Raider. Here 7950X with its 16 cores is able to load up the GPU with a whole lot of frames which makes it lead from 12900K by 20% on average FPS and 6% on 1 percentiles. Important thing to note - the performance per watt is actually even between the two.
Next is Fortnite and here Ryzen 7 gets out in the lead followed by Intel 12900K and only then is the new Ryzen 9. The average frame difference is around 6% between each of the devices. Ryzen 9 is likely much lower due to having multiple chiplets as compared to Ryzen 7 with single chiplet design.
In Horizon Zero Dawn we see Intel taking the lead and that is especially visible on the 1 percentile performance which is about 7% faster than 7700X. In this example there is actually very little improvement between 5950X and 7950X besides 6% on the 1 percentiles.
And for the last game we have Overwatch where results are very much neck to neck. But there is a really good reason for this. Overwatch has 400 FPS in game cap which makes it very difficult to do reasonable CPU benchmarking at the high end.
This is a two fold issue - the fast CPUs we are testing here are pushing crazy framerates that get past 400 - to reduce the framerate we need to increase the resolution. This then makes GPU the bottleneck and we are stuck with this kind of worthless result. If someone knows how I can remove the 400 FPS limit, please leave the info in the comments below..
One new addition to the Ryzen 7000 series is integrated Radeon GPU so we tested these out. They are great for troubleshooting, especially if you are overclocking. Let us know if you would like to see us do an overclocking guide.
Also you could technically do some light gaming. Check out this performance in Fortnite. Even though this is 1080p low, it still delivers over 100 average FPS and 71 FPS on 1 percentiles, so if you are in a pinch or waiting for the new graphics card then this may have some value to you.
Performance in Overwatch is not as good though. Here Intel barely hits the 60 average FPS and AMD doesn't even hit it at all with pretty bad 1 percentile performance on the 7950x.
This brings us really well to the conclusion - the Ryzen 9 7950X is clearly a very powerful CPU and comes with generational improvements in performance as well as some new features. At 699 USD it is a premium chip but costs don’t end here. The AM5 motherboards aren't exactly cheap and the RAM that we used in this build is 280 USD for the kit. All of this together makes it a very expensive build, but at the same time for a production environment that is justifiable.
If on the other hand you are looking to build a gaming PC, I would actually look closer at Ryzen 7 7700X and put the extra 300 USD towards other components. Essentially, for productivity at this price point there is currently no competition and I am doubtful that Intel will have anything to answer with, but we will just have to wait and see.