Wasted Opportunity: AMD Ryzen 7 9700X CPU Review & Benchmarks vs. 7800X3D, 7700X, & More
video description
Date: 2024-08-07
Related videos
Comments and reviews: 20
photonboy999
STEVE... my only takeaway from this video is I should buy the Intel i9-14900K
(I may have misunderstood... bad joke aside, I'm still rocking an R9-3900x. I upgraded to an RTX4070 to replace my GTX1080, then got a 4K 144Hz GSync monitor. CPU bottleneck in a few games, but meh, doesn't matter nearly so much if I drop below 60FPS any more. BEFORE I got stutters at times due to VSYNC issues. Now I'm pretty dang happy (even figured out how to run modded SKYRIM AE at a locked 90FPS/4K). Is this a balanced system now Actually, YES. Despite the Ryzen 3000 CPU being a bit old, MOST games are still bottlenecked by my graphics card. I have NO reason to upgrade my computer any time soon... why am I here watching CPU videos Dunno...I figure if I hold onto my 12-core CPU long enough games will start using more than 8 of them (more than 67%) and justify the purpose. (got it for Handbrake actually))
reply
STEVE... my only takeaway from this video is I should buy the Intel i9-14900K
(I may have misunderstood... bad joke aside, I'm still rocking an R9-3900x. I upgraded to an RTX4070 to replace my GTX1080, then got a 4K 144Hz GSync monitor. CPU bottleneck in a few games, but meh, doesn't matter nearly so much if I drop below 60FPS any more. BEFORE I got stutters at times due to VSYNC issues. Now I'm pretty dang happy (even figured out how to run modded SKYRIM AE at a locked 90FPS/4K). Is this a balanced system now Actually, YES. Despite the Ryzen 3000 CPU being a bit old, MOST games are still bottlenecked by my graphics card. I have NO reason to upgrade my computer any time soon... why am I here watching CPU videos Dunno...I figure if I hold onto my 12-core CPU long enough games will start using more than 8 of them (more than 67%) and justify the purpose. (got it for Handbrake actually))
reply
kylehutto8044
My reading of the tea leaves is that Im sensing that in the near future, AMD is gonna split the X3D parts and Ryzen. Maybe bring back the FX branding. Ryzen will be the datacenter/workstation/home office focused CPU and X3D/FX will be the gaming focused line entirely. As in AMD will stop targeting single core performance at all on the regular Ryzen line and it will be entirely IPC/watt gain focused, whereas the X3D will be focused on 6-12 cores at the highest possible IPC and frequency and however much cache they can stuff on it.
If this is the beginning of that split, this launch makes sense. AMD is no longer targeting gamers with regular Ryzen. Everything about this is stuff that will excite datacenter mangers and system integrators. The only missing piece of the puzzle is the direction they will take on laptops with no X3D equivalent currently.
reply
My reading of the tea leaves is that Im sensing that in the near future, AMD is gonna split the X3D parts and Ryzen. Maybe bring back the FX branding. Ryzen will be the datacenter/workstation/home office focused CPU and X3D/FX will be the gaming focused line entirely. As in AMD will stop targeting single core performance at all on the regular Ryzen line and it will be entirely IPC/watt gain focused, whereas the X3D will be focused on 6-12 cores at the highest possible IPC and frequency and however much cache they can stuff on it.
If this is the beginning of that split, this launch makes sense. AMD is no longer targeting gamers with regular Ryzen. Everything about this is stuff that will excite datacenter mangers and system integrators. The only missing piece of the puzzle is the direction they will take on laptops with no X3D equivalent currently.
reply
gamersnexus
I mean, clearly its not operating properly. You should have just cancelled both reviews and told us why. These results clearly don't represent the product people will be buying and Linus easily got 5.2-5.3GHz Boost in All-Core loads.
Im not saying AMD is excused for giving reviewers clearly defective chips, but whats the point of a review with a broken product
Linus saw 10-20% uplift in basically every single workload versus the 7700X and sometimes even bigger gains by manually turning up the power limit to match the 7700X's power consumption.
Its not the most fantastic uplift after 2 full years but its certainly decent and Ryzen 5000 series owners will certainly be considering upgrades when they see the real performance on display.
reply
I mean, clearly its not operating properly. You should have just cancelled both reviews and told us why. These results clearly don't represent the product people will be buying and Linus easily got 5.2-5.3GHz Boost in All-Core loads.
Im not saying AMD is excused for giving reviewers clearly defective chips, but whats the point of a review with a broken product
Linus saw 10-20% uplift in basically every single workload versus the 7700X and sometimes even bigger gains by manually turning up the power limit to match the 7700X's power consumption.
Its not the most fantastic uplift after 2 full years but its certainly decent and Ryzen 5000 series owners will certainly be considering upgrades when they see the real performance on display.
reply
JJFX-
Well, never underestimate AMD's ability to shoot themselves in the foot. They should have just waited to release the whole stack together. Sending you of all people a defective chip is just mind-blowing.
There's something strange about this generation... I thought they were sand-bagging but it seems they might really be struggling with clocks. I'm sure compatibility will get resolved but it's looking like memory speeds really won't be much different than Zen4, which isn't surprising.
I'm just curious to see these tested with higher power limits. There must be a reason AMD choked these off to the point of effectively matching performance of the previous gen. The dual CCD performance will shine more light on what's likely going on.
reply
Well, never underestimate AMD's ability to shoot themselves in the foot. They should have just waited to release the whole stack together. Sending you of all people a defective chip is just mind-blowing.
There's something strange about this generation... I thought they were sand-bagging but it seems they might really be struggling with clocks. I'm sure compatibility will get resolved but it's looking like memory speeds really won't be much different than Zen4, which isn't surprising.
I'm just curious to see these tested with higher power limits. There must be a reason AMD choked these off to the point of effectively matching performance of the previous gen. The dual CCD performance will shine more light on what's likely going on.
reply
eliesercepeda686
Intel is asking me to pay them 700 dollars upfront (the cost of a new 14900KS) before they even send me a replacement. Then will refund me after they received the bad one, since I’m overseas it can take a month for them to get that bad one. Oh and the cost of shipping is included on that as well. Otherwise ship your CPU first wait a month until they get it, then wait a month until they ship it to me. Imagine waiting that long without a working computer until you get your replacement. I have NEVER dealt with a business that asked for money UPFRONT BEFORE they shipped you a replacement product for an already damaged product by their own design lol why are they stressing if they KNOW it’s a bad CPU asking for 700 dollars
reply
Intel is asking me to pay them 700 dollars upfront (the cost of a new 14900KS) before they even send me a replacement. Then will refund me after they received the bad one, since I’m overseas it can take a month for them to get that bad one. Oh and the cost of shipping is included on that as well. Otherwise ship your CPU first wait a month until they get it, then wait a month until they ship it to me. Imagine waiting that long without a working computer until you get your replacement. I have NEVER dealt with a business that asked for money UPFRONT BEFORE they shipped you a replacement product for an already damaged product by their own design lol why are they stressing if they KNOW it’s a bad CPU asking for 700 dollars
reply
powerpower-rg7bk
This feels like a non-X part given the focus on efficiency. I'm curious if given the same power and limits as the 7700X that this would pull ahead. This feels like AMD was playing things too safe.
Speaking of power limits, are there any overclocking results The silicon has similar stock turbos so even just bringing out both the 7700X and 9700X to those same clocks but for an all core baseline, I'd hope to see the IPC gains shine while hopefully consuming less power.
The LAMMPS test result is really really interesting. My presumption was that this was a heavy AVX-512 workload which is one area where this chip should really shine, not a regression in performance.
reply
This feels like a non-X part given the focus on efficiency. I'm curious if given the same power and limits as the 7700X that this would pull ahead. This feels like AMD was playing things too safe.
Speaking of power limits, are there any overclocking results The silicon has similar stock turbos so even just bringing out both the 7700X and 9700X to those same clocks but for an all core baseline, I'd hope to see the IPC gains shine while hopefully consuming less power.
The LAMMPS test result is really really interesting. My presumption was that this was a heavy AVX-512 workload which is one area where this chip should really shine, not a regression in performance.
reply
AndreGreeff
interesting results indeed, and definitely very disappointing... I built a system at the start of this year, X670-E platform with a R5 7600X CPU and a 7800XT GPU, intending to upgrade the CPU when these Zen 5 chips were released, but now I'm starting to think my current system is going to stay the way it is for a while longer.. I do still love seeing my 7600X sitting so consistently in the middle of all these tests. never the best, never the worst, but it really is a good all-round chip.
also, completely random: in regards to CPU model numbers, Steve pronounced the letter X 183 times in this video.. including X, X3D and non-X phrases. (:
reply
interesting results indeed, and definitely very disappointing... I built a system at the start of this year, X670-E platform with a R5 7600X CPU and a 7800XT GPU, intending to upgrade the CPU when these Zen 5 chips were released, but now I'm starting to think my current system is going to stay the way it is for a while longer.. I do still love seeing my 7600X sitting so consistently in the middle of all these tests. never the best, never the worst, but it really is a good all-round chip.
also, completely random: in regards to CPU model numbers, Steve pronounced the letter X 183 times in this video.. including X, X3D and non-X phrases. (:
reply
woobilicious.
Seems like a poor choice to reduce these to 65w chips, Zen 5 is an 15% IPC improvement, but them being throttled heavily in multicore (by -16% clock) seems like it completely erases these improvements, There are hints these are stellar parts with PBO if you do the math, I suspect the they're either dropping some XT parts at 105w or trying to position the 9900X as the best option.
I do wonder if there's some meta game here where if they get enough people using 12C/24T parts, that Intel's big.little design starts to be the worst of both worlds, where games with 12 threads will just cry trying to run the 2-4 extra threads on E cores.
reply
Seems like a poor choice to reduce these to 65w chips, Zen 5 is an 15% IPC improvement, but them being throttled heavily in multicore (by -16% clock) seems like it completely erases these improvements, There are hints these are stellar parts with PBO if you do the math, I suspect the they're either dropping some XT parts at 105w or trying to position the 9900X as the best option.
I do wonder if there's some meta game here where if they get enough people using 12C/24T parts, that Intel's big.little design starts to be the worst of both worlds, where games with 12 threads will just cry trying to run the 2-4 extra threads on E cores.
reply
novantha1
While this review didn't really cover it, I'm pretty sure Zen 5 is suppose to be an absolute monster in AVX workloads (particularly AI workloads that leverage VNNI) which I do pretty regularly, and I'm in a situation where both absolute power consumption and power efficiency are a huge concern, so I am unbelievably hyped for this gen, though I do understand that this is probably a lot less exciting for the average gamer who maybe just wants a crazy good CPU at whatever the power cost is.
I will say that there will probably be a lot of headroom to overclock the lower TDP Ryzen 9000 parts, IMO.
reply
While this review didn't really cover it, I'm pretty sure Zen 5 is suppose to be an absolute monster in AVX workloads (particularly AI workloads that leverage VNNI) which I do pretty regularly, and I'm in a situation where both absolute power consumption and power efficiency are a huge concern, so I am unbelievably hyped for this gen, though I do understand that this is probably a lot less exciting for the average gamer who maybe just wants a crazy good CPU at whatever the power cost is.
I will say that there will probably be a lot of headroom to overclock the lower TDP Ryzen 9000 parts, IMO.
reply
narayanvarma3311
9th gen Ryzen should have been a laptop only launch, they could have continued selling 7th gen for a year or more Intel still wouldn't have caught up for gaming. It is purely a launch for the investors.
They could have upped the core count for the parts that were losing to Intel in productivity but they didn't care enough to make those changes this gen. If they had pushed that envelope and sorted out stability, this gen would have been meaningful.
AMD as always, is great at grabbing mediocrity whenever it has a chance at really pressing their advantage.
reply
9th gen Ryzen should have been a laptop only launch, they could have continued selling 7th gen for a year or more Intel still wouldn't have caught up for gaming. It is purely a launch for the investors.
They could have upped the core count for the parts that were losing to Intel in productivity but they didn't care enough to make those changes this gen. If they had pushed that envelope and sorted out stability, this gen would have been meaningful.
AMD as always, is great at grabbing mediocrity whenever it has a chance at really pressing their advantage.
reply
laurenz4572
I think you need to retest with PBO to get the proper performance uplift. Seems to me AMD is catering to the efficiency crowd with the out-of-box settings and to contrast themselves from Intel, people looking for significant performance uplift will have to activate PBO. All Zen 5 parts buts especially the 9700X seem to be held back by default power limits, judging from the slides AMD presented the 9700X will benefit most from PBO. The mixed/ bad reviews shouldn't be a surprise really, what did AMD expect when they gave the 700 part the same 65W TDP as the 600 part.
reply
I think you need to retest with PBO to get the proper performance uplift. Seems to me AMD is catering to the efficiency crowd with the out-of-box settings and to contrast themselves from Intel, people looking for significant performance uplift will have to activate PBO. All Zen 5 parts buts especially the 9700X seem to be held back by default power limits, judging from the slides AMD presented the 9700X will benefit most from PBO. The mixed/ bad reviews shouldn't be a surprise really, what did AMD expect when they gave the 700 part the same 65W TDP as the 600 part.
reply
apowerr
It’s important to remember that x86_64 CPU architectures like Zen5 also compete with incredibly efficient ARM64 products such as Apple’s M series.
Alongside these gaming and enthusiast desktop CPUs, the Zen5 architecture will power all sorts of laptops, tablets, handhelds, all-in-one business and consumer machines and the next generation of the EPYC server architecture.
Focusing this generation on such significant gains in efficiency, power draw and thermals while sticking to only having Pcores is a massive win for AMD and X86_64 computing as we know it.
reply
It’s important to remember that x86_64 CPU architectures like Zen5 also compete with incredibly efficient ARM64 products such as Apple’s M series.
Alongside these gaming and enthusiast desktop CPUs, the Zen5 architecture will power all sorts of laptops, tablets, handhelds, all-in-one business and consumer machines and the next generation of the EPYC server architecture.
Focusing this generation on such significant gains in efficiency, power draw and thermals while sticking to only having Pcores is a massive win for AMD and X86_64 computing as we know it.
reply
user-dm7sk5wf5w
I would have loved to see an overclocked 9700x in the chart. What could have amd done with all that power savings if it allowed the cpu to consume alll that saved power. I feel like the review seems to almost not care(the almost is because there is that one efficiency chart) about the enormous power difference in these cpus. Its like cooler testing and noise normalizing. Point is your conclusion almost doesn't seem swayed by the fact that the 7700x is using 168.4% of the power of the 9700x. That's extremely impressive for the 9700x, yet is given a meh.
reply
I would have loved to see an overclocked 9700x in the chart. What could have amd done with all that power savings if it allowed the cpu to consume alll that saved power. I feel like the review seems to almost not care(the almost is because there is that one efficiency chart) about the enormous power difference in these cpus. Its like cooler testing and noise normalizing. Point is your conclusion almost doesn't seem swayed by the fact that the 7700x is using 168.4% of the power of the 9700x. That's extremely impressive for the 9700x, yet is given a meh.
reply
4nyNoob
since ryzen 5000 AMD didn't launch any low end chips, I've been watching since the beginning, and the 2200G brought me a lot of joy
i want to build amd PCs for my friends because i know that the socket will be well lived, but the lack of low end parts on Ryzen makes it a pretty steep entry
the 9700x is basically a cooler 7700x, and even with a price cut i do not see this as a good successor, i don't want to see ryzen pull a skylake stagnation, if this performance level is a BIOS problem, then a bigger delay would be preferable to a meh launch
reply
since ryzen 5000 AMD didn't launch any low end chips, I've been watching since the beginning, and the 2200G brought me a lot of joy
i want to build amd PCs for my friends because i know that the socket will be well lived, but the lack of low end parts on Ryzen makes it a pretty steep entry
the 9700x is basically a cooler 7700x, and even with a price cut i do not see this as a good successor, i don't want to see ryzen pull a skylake stagnation, if this performance level is a BIOS problem, then a bigger delay would be preferable to a meh launch
reply
hideinbush1133
Maybe you mentioned this and I missed it, but equivalent performance for 30% more efficiency seems like a shift in focus towards server or industrial applications. Couple this with companies saying they want to transition their server processors away from Intel, and the Semi organization market forecast saying the server market is going to be the greatest area of growth for 2025, I think these processors are very well positioned. It's just that gaming household consumers aren't the target audience, especially for the non X3D variants.
reply
Maybe you mentioned this and I missed it, but equivalent performance for 30% more efficiency seems like a shift in focus towards server or industrial applications. Couple this with companies saying they want to transition their server processors away from Intel, and the Semi organization market forecast saying the server market is going to be the greatest area of growth for 2025, I think these processors are very well positioned. It's just that gaming household consumers aren't the target audience, especially for the non X3D variants.
reply
p4radigm989
It's hard to please Gaymers.
first they complain that CPUs run at 90c, then we get Intel CPUs degrading themselves by using too much Power, but when AMD releases a CPU that runs 20c cooler and uses way less energy, it's suddenly NOT EXCITING. Personally I think having to pay a lot less on monthly Energy bills is very exciting, as is having a CPU that doesn't burn itself out at 90c. Gaymers who get more excited by an extra 50fps when they're already in the 100s, feel free to overvolt your CPU. After all, that's fun too.
reply
It's hard to please Gaymers.
first they complain that CPUs run at 90c, then we get Intel CPUs degrading themselves by using too much Power, but when AMD releases a CPU that runs 20c cooler and uses way less energy, it's suddenly NOT EXCITING. Personally I think having to pay a lot less on monthly Energy bills is very exciting, as is having a CPU that doesn't burn itself out at 90c. Gaymers who get more excited by an extra 50fps when they're already in the 100s, feel free to overvolt your CPU. After all, that's fun too.
reply
kingcrimson5483
I think the dismal gains we see here are probably due to the conservative power targets that amd set. Ipc wise zen 5 seems considerably better than zen 4, being 10-26% better depending on task. Running PBO on both zen 4 and 5 shows a pretty significant gain on zen 5.
It seems amd has lost the multicore crown to intel, while having the best single core perf for a zen CPU. Maybe theyll add a 9800C for multicore.
They're probably keeping these chips low power for more product segmentation.
reply
I think the dismal gains we see here are probably due to the conservative power targets that amd set. Ipc wise zen 5 seems considerably better than zen 4, being 10-26% better depending on task. Running PBO on both zen 4 and 5 shows a pretty significant gain on zen 5.
It seems amd has lost the multicore crown to intel, while having the best single core perf for a zen CPU. Maybe theyll add a 9800C for multicore.
They're probably keeping these chips low power for more product segmentation.
reply
themobster7284
Dont see a single reason to sell my 2017 AM4 x370 high end mobo, and 5800x3D, and to jump on AM5, will prob wait for AM6 or even AM7, and prices did gone so much up... And another thing back in AM4 days, AMD had only BIOS problems and memory compatibility problems and low cpu and ram clock speeds, but it was 110% HW problem free.... These days since AM5 did launch they got also HW problems, burned out CPUs sockets, and now this crap with Zen V, seems the QC did gone to hell these days...
reply
Dont see a single reason to sell my 2017 AM4 x370 high end mobo, and 5800x3D, and to jump on AM5, will prob wait for AM6 or even AM7, and prices did gone so much up... And another thing back in AM4 days, AMD had only BIOS problems and memory compatibility problems and low cpu and ram clock speeds, but it was 110% HW problem free.... These days since AM5 did launch they got also HW problems, burned out CPUs sockets, and now this crap with Zen V, seems the QC did gone to hell these days...
reply
snakeat3r114
We are living in an age where we are very close to the limits of the x86/64 processors. That's why intel failed to improve from the 13th gen and now we are seeing same thing with AMD. The most advancement that we see today is because of memory speed. Long gone are the 50% generational performance increase. The truth is if you have a chip like 5800x3d you will probably still be able to run the latest games like 8-10 years from now, unless the core count starts to matter a lot.
reply
We are living in an age where we are very close to the limits of the x86/64 processors. That's why intel failed to improve from the 13th gen and now we are seeing same thing with AMD. The most advancement that we see today is because of memory speed. Long gone are the 50% generational performance increase. The truth is if you have a chip like 5800x3d you will probably still be able to run the latest games like 8-10 years from now, unless the core count starts to matter a lot.
reply
B.Bogdan
Couldn't the same thing that you showed happens in Blender (bad all-core boost) happen in the rest of the tests Like the boosting algorithm is out of whack It seems like the Blender scores are similar to the other tests (small differences between 9700x and 7700x), considering that the working clocks are way lower on the 9700X in blender.
Maybe try plotting the differences between 7700 and 9700 core clocks during 2-3 games would show this
reply
Couldn't the same thing that you showed happens in Blender (bad all-core boost) happen in the rest of the tests Like the boosting algorithm is out of whack It seems like the Blender scores are similar to the other tests (small differences between 9700x and 7700x), considering that the working clocks are way lower on the 9700X in blender.
Maybe try plotting the differences between 7700 and 9700 core clocks during 2-3 games would show this
reply
Add a review, comment
Other channel videos
