Crazy Good Efficiency: AMD Ryzen 9 7900 CPU Benchmarks & Thermals
video description
Date: 2023-01-11
Related videos
Comments and reviews: 14
Hardcore
Gamer's Nexus Me and my friends aren't getting advertised clock speed boost on our 7950x's and we cannot even replicate the 38k R23 scores of reviewers nor are our systems stable with curb optimizer settings. We have tried hours upon hours to get curb settings stable and any form of ECO settings results in crashes. My friend can't even manage above 33k in R23 at stock settings and it runs 95 c with a 420mm AIO but I can manage 37k with a static 5.0ghz 1.150v OC but at stock under 35k. Boost clocks in game are around 4.6ghz with a peak of 5.0ghz on one core for a few seconds, it mainly sits on 4.6ghz to 4.8ghz.
We tried latest bios and rolling back to older bios but still the same stability issues, 3x of us have spent hours on attempting to get curb settings stable when not even -10 is stable on either of our chips. In one scenario -20 is stable then in another we get crashes, sometimes I can't even get Premier or Vegas to open and the constant crashes corrupted various installs of applications which was really annoying.
My question is how did reviewers get 38k in R23? How are reviewers getting -30 Curb stable while we cannot even manage -10? How are boosts speeds so much better in reviewers results and we can't even manage 5.0ghz in games with it holding boost? Could it be binned samples? Only thing that would explain this.
We are now considering cutting our losses and taking the loss to sell them off and buying 13900k's instead but trouble is hardly anybody wants AM5, its to expensive and runs to hot. I ordered a 360mm AIO from EK so I'm hoping it reduces temps of my GPU because the added heat in my 802 from the 7950x is causing my FTW3 Ultra 3090 to throttle down to under 1800mhz and run at 83 c 100% fan speed. The extra heat this system puts out makes me sweat so much my headset gets drenched while playing a game or even just sitting near it at idle (Australia, its hot here) but my friend is in Austria where it is cold and with a 420mm AIO it still runs hot.
We aren't getting advertised performance, I've left numerous comments on reviewers videos with none of them responding so I would appreciate a response from Gamer's Nexus regarding our issues.
Its getting really hard to make purchasing decisions based on reviews anymore since the products us consumers buy do not perform the same as the samples reviewers get. Imo reviews need to be done on retail samples over manufacturer provided samples.
We are quite disappointed with our 7950x's and after spending hours upon hours we are tired of messing around and I'm tired of having the Sun in my 802 minus the power of it.
I would appreciate your attention on this matter because we are stumped, thank you Steve and GN team.
-Asrock Taichi Carrara x670-E
-Kingston Fury Renegade CL32 6000mhz 2x16gb
reply
Gamer's Nexus Me and my friends aren't getting advertised clock speed boost on our 7950x's and we cannot even replicate the 38k R23 scores of reviewers nor are our systems stable with curb optimizer settings. We have tried hours upon hours to get curb settings stable and any form of ECO settings results in crashes. My friend can't even manage above 33k in R23 at stock settings and it runs 95 c with a 420mm AIO but I can manage 37k with a static 5.0ghz 1.150v OC but at stock under 35k. Boost clocks in game are around 4.6ghz with a peak of 5.0ghz on one core for a few seconds, it mainly sits on 4.6ghz to 4.8ghz.
We tried latest bios and rolling back to older bios but still the same stability issues, 3x of us have spent hours on attempting to get curb settings stable when not even -10 is stable on either of our chips. In one scenario -20 is stable then in another we get crashes, sometimes I can't even get Premier or Vegas to open and the constant crashes corrupted various installs of applications which was really annoying.
My question is how did reviewers get 38k in R23? How are reviewers getting -30 Curb stable while we cannot even manage -10? How are boosts speeds so much better in reviewers results and we can't even manage 5.0ghz in games with it holding boost? Could it be binned samples? Only thing that would explain this.
We are now considering cutting our losses and taking the loss to sell them off and buying 13900k's instead but trouble is hardly anybody wants AM5, its to expensive and runs to hot. I ordered a 360mm AIO from EK so I'm hoping it reduces temps of my GPU because the added heat in my 802 from the 7950x is causing my FTW3 Ultra 3090 to throttle down to under 1800mhz and run at 83 c 100% fan speed. The extra heat this system puts out makes me sweat so much my headset gets drenched while playing a game or even just sitting near it at idle (Australia, its hot here) but my friend is in Austria where it is cold and with a 420mm AIO it still runs hot.
We aren't getting advertised performance, I've left numerous comments on reviewers videos with none of them responding so I would appreciate a response from Gamer's Nexus regarding our issues.
Its getting really hard to make purchasing decisions based on reviews anymore since the products us consumers buy do not perform the same as the samples reviewers get. Imo reviews need to be done on retail samples over manufacturer provided samples.
We are quite disappointed with our 7950x's and after spending hours upon hours we are tired of messing around and I'm tired of having the Sun in my 802 minus the power of it.
I would appreciate your attention on this matter because we are stumped, thank you Steve and GN team.
-Asrock Taichi Carrara x670-E
-Kingston Fury Renegade CL32 6000mhz 2x16gb
reply
fatih
I miss the zen 2 launch era. Pricing was good, performance was amazing, efficiency was great. AMD did really win with zen 2 launch and also zen 3 without a question. AMD forced intel to start working again with their zen 2 and that's why today we have the amazing 12th gen and 13th gen intel cpu series. But AMD for the last year or so literally f cked up with their pricing strategy and also they lost the performance crown, they lost the price/performance crown too and at least the non x series can still hold the efficiency crown. But they still don't have anything against intel's mid-range stack in terms of both price and performance. AMD needs to heavily discount for the 7000 series or they will not sell or I don't know maybe repurpose those non x cpu's as laptop cpu's. That'd make sense for high-end notebooks.
reply
I miss the zen 2 launch era. Pricing was good, performance was amazing, efficiency was great. AMD did really win with zen 2 launch and also zen 3 without a question. AMD forced intel to start working again with their zen 2 and that's why today we have the amazing 12th gen and 13th gen intel cpu series. But AMD for the last year or so literally f cked up with their pricing strategy and also they lost the performance crown, they lost the price/performance crown too and at least the non x series can still hold the efficiency crown. But they still don't have anything against intel's mid-range stack in terms of both price and performance. AMD needs to heavily discount for the 7000 series or they will not sell or I don't know maybe repurpose those non x cpu's as laptop cpu's. That'd make sense for high-end notebooks.
reply
Earth
The 7950X with a PBO tuned, negative curve optimiser you can get it to boost to 5.7 Ghz at a lower CPU Package Power. To better utilise the negative curve optimiser set a static voltage of 1.3 volts. I have mine boosting to 5.5 Ghz all core with PBO tuned, Thermal limited to 80 degrees c and using 160 watts - no matter the workload or duration CCD1 never exceeds 87 degrees and CCD2 never exceeds 82 degrees using a arctic liquid freezer 2 360. Undervolting a must with AM5 as there is noticeable degradation as on stock 1.5 volts is shoved down each core. Before all my cores could boost to 5.7 Ghz out of the box, after prolonged testing on stock to isolate memory issues now only some of my cores can boost to 5.7 Ghz . Anyway, using similar settings on the 7900 should yield alot more better performance.
reply
The 7950X with a PBO tuned, negative curve optimiser you can get it to boost to 5.7 Ghz at a lower CPU Package Power. To better utilise the negative curve optimiser set a static voltage of 1.3 volts. I have mine boosting to 5.5 Ghz all core with PBO tuned, Thermal limited to 80 degrees c and using 160 watts - no matter the workload or duration CCD1 never exceeds 87 degrees and CCD2 never exceeds 82 degrees using a arctic liquid freezer 2 360. Undervolting a must with AM5 as there is noticeable degradation as on stock 1.5 volts is shoved down each core. Before all my cores could boost to 5.7 Ghz out of the box, after prolonged testing on stock to isolate memory issues now only some of my cores can boost to 5.7 Ghz . Anyway, using similar settings on the 7900 should yield alot more better performance.
reply
PaladinVII
Glad that the tests were consistent. I ordered the Ryzen 9 7900 (no-x) the moment it launched, using the 7600 and 7700 reviews (and LTT's 7900 review) as guidelines. I saw the pattern of 88w and equal enough performance to its X-brother. My computer is over 6 years old. The 1080 that was in it already died and I'm using a friend's 1060. Anything I bought at this point is going to be about 5x the speed of what I had when it was working. Thank you GN for these videos. They have been critical to my decision making in building my new computer.
reply
Glad that the tests were consistent. I ordered the Ryzen 9 7900 (no-x) the moment it launched, using the 7600 and 7700 reviews (and LTT's 7900 review) as guidelines. I saw the pattern of 88w and equal enough performance to its X-brother. My computer is over 6 years old. The 1080 that was in it already died and I'm using a friend's 1060. Anything I bought at this point is going to be about 5x the speed of what I had when it was working. Thank you GN for these videos. They have been critical to my decision making in building my new computer.
reply
3XOUT
About time thermals were even mentioned again. I still do not get why you guys don't include it generally. I can't be the only one interested. Most ppl don't like their fans spinning at 100% constantly (like you do in your testing), you do know that right? ;)
It's even more relevant for the average consumer who never tweaks their CPU and just buys a system that, for some reason, sounds like a jet engine all of the time.
But, what did the PBO tweak do to the impressive thermals? Didn't hear you mention that, but might have missed it.
reply
About time thermals were even mentioned again. I still do not get why you guys don't include it generally. I can't be the only one interested. Most ppl don't like their fans spinning at 100% constantly (like you do in your testing), you do know that right? ;)
It's even more relevant for the average consumer who never tweaks their CPU and just buys a system that, for some reason, sounds like a jet engine all of the time.
But, what did the PBO tweak do to the impressive thermals? Didn't hear you mention that, but might have missed it.
reply
Forthereg
All the production benchmarks are fantastic. But you can clearly see the gaming benchmarks are just silly and pointless at this point 500+fps . (with even 12100F pulling 120fps).
It would have been really nice if you had actual meaningful games where CPU mattered. IE simulation games.
like Cities skyline, Factorio, Oxygen not included . All these games that just CHOKE the living hell out of CPU once you get into them deep.
P.S I play these 3 games at 20 FPS on a 11700K. Thanks Steve !
reply
All the production benchmarks are fantastic. But you can clearly see the gaming benchmarks are just silly and pointless at this point 500+fps . (with even 12100F pulling 120fps).
It would have been really nice if you had actual meaningful games where CPU mattered. IE simulation games.
like Cities skyline, Factorio, Oxygen not included . All these games that just CHOKE the living hell out of CPU once you get into them deep.
P.S I play these 3 games at 20 FPS on a 11700K. Thanks Steve !
reply
Shino
It can be a lot more useful if AMD would provide a way to control PBO inside windows other than ryzen master. They provide a ryzen master monitoring SDK but it is only for monitoring and can't be used to control PBO inside other applications. Sometimes the amount of power is available changes during the day (like any application that is depending on solar power) and having a way to change PBO behavior inside other applications can be really helpful.
reply
It can be a lot more useful if AMD would provide a way to control PBO inside windows other than ryzen master. They provide a ryzen master monitoring SDK but it is only for monitoring and can't be used to control PBO inside other applications. Sometimes the amount of power is available changes during the day (like any application that is depending on solar power) and having a way to change PBO behavior inside other applications can be really helpful.
reply
Vedthrfolnir
Can you please compare like-to-like and turn on PBO on the X SKUs too? It feels disingenuous to compare the non-X PBO to the X non-PBO and say look they are the same, obviously by the cheaper one . Does the X SKU not also gain performance from PBO? Even if that's true, it would be helpful to show that. Especially since PBO is so easy to use, basically everyone should be turning it on unless they have a niche reason not to.
reply
Can you please compare like-to-like and turn on PBO on the X SKUs too? It feels disingenuous to compare the non-X PBO to the X non-PBO and say look they are the same, obviously by the cheaper one . Does the X SKU not also gain performance from PBO? Even if that's true, it would be helpful to show that. Especially since PBO is so easy to use, basically everyone should be turning it on unless they have a niche reason not to.
reply
SharkBender
Shit in stil not sure of my purchase. I game on a 1440p monitor with a Ryzen 5 3600 and a RX6800. I now ordered a Ryzen 7 5700x for 209 euro. Is this a good choice for an upgrade or did I need to go for the 5800x3D/5900x? I only game on my pc
5800x3D = 400
5900x = 380
I hope I made the wise decision
Games I play:
Tarkov, MW2, Valheim, Elden Ring, Battlefield, WoW, Hell let Loose, Squat.
reply
Shit in stil not sure of my purchase. I game on a 1440p monitor with a Ryzen 5 3600 and a RX6800. I now ordered a Ryzen 7 5700x for 209 euro. Is this a good choice for an upgrade or did I need to go for the 5800x3D/5900x? I only game on my pc
5800x3D = 400
5900x = 380
I hope I made the wise decision
Games I play:
Tarkov, MW2, Valheim, Elden Ring, Battlefield, WoW, Hell let Loose, Squat.
reply
Quiet
I m not an electrical engineer or anything so can someone please explain the power consumption and efficiency results? If the thing uses 86.4 watts doing blender for 5 minutes, how does that translate into 13.8 watt hours?? That doesn t make any sense to me.
Honestly, I just wanna know how much electricity the thing uses for high end games per hour so I can figure out roughly how expensive it is to run.
reply
I m not an electrical engineer or anything so can someone please explain the power consumption and efficiency results? If the thing uses 86.4 watts doing blender for 5 minutes, how does that translate into 13.8 watt hours?? That doesn t make any sense to me.
Honestly, I just wanna know how much electricity the thing uses for high end games per hour so I can figure out roughly how expensive it is to run.
reply
Templar
This is why we definitely should encourage these beloved corporations to include more Xs in product names.
One X doesn't make any difference in performance and it doesn't make sense either.
If I were AMD, I'd responsibly introduce Ryzen 9 7900 XXX edition later on at the 7900X price range so it will be more sensibly justifying the price of the performance you'll get.
reply
This is why we definitely should encourage these beloved corporations to include more Xs in product names.
One X doesn't make any difference in performance and it doesn't make sense either.
If I were AMD, I'd responsibly introduce Ryzen 9 7900 XXX edition later on at the 7900X price range so it will be more sensibly justifying the price of the performance you'll get.
reply
Temporal
Man, the way the 13900K runs away with a lead on some of the benchmarks reminds me of the way the RTX 4090 does the same on the GPU side. I imagine pairing the two and using a 4K monitor is an incredible gaming/productivity experience. Using a 1440p monitor would be a game of find the highest refresh rate on the market and basically forget that graphics settings exist.
reply
Man, the way the 13900K runs away with a lead on some of the benchmarks reminds me of the way the RTX 4090 does the same on the GPU side. I imagine pairing the two and using a 4K monitor is an incredible gaming/productivity experience. Using a 1440p monitor would be a game of find the highest refresh rate on the market and basically forget that graphics settings exist.
reply
Hiranya
Honestly, if we are primarily comparing 7900 to 7900X, I would've liked to see 7900X PBO results in the charts as well. Because although I get the point you can get 7900X performance by just putting 7900 on PBO , I don't think it's completely fair without including 7900X PBO as well. Not sure why that isn't included, is there something obvious I'm missing?
reply
Honestly, if we are primarily comparing 7900 to 7900X, I would've liked to see 7900X PBO results in the charts as well. Because although I get the point you can get 7900X performance by just putting 7900 on PBO , I don't think it's completely fair without including 7900X PBO as well. Not sure why that isn't included, is there something obvious I'm missing?
reply
Mustafa
With honestly the kind of amazing performance of the non x 7000 series CPUs, I wonder if silicone lottery will eventually become a bigger factor in the future where these CPUs don t all actually perform this well when power unlocked. I m not sure if these are all review units but even if they aren t, maybe the first batch is on better silicone?
reply
With honestly the kind of amazing performance of the non x 7000 series CPUs, I wonder if silicone lottery will eventually become a bigger factor in the future where these CPUs don t all actually perform this well when power unlocked. I m not sure if these are all review units but even if they aren t, maybe the first batch is on better silicone?
reply
Add a review, comment
Other channel videos
