Thanks for the review. I personally would have liked to see a 7800X3D in the benchmarks. Also, I would have omitted 720p and replaced it with 1440p gaming benchmarks. 1080p is perfectly fine for a synthetic benchmark to see which CPU is faster overall and a lot of gamers run 1440p.
Agreed. The 7800X3D should be very close in price to the 9700X so having both in the gaming charts would be pretty nice. Offhand I assume it will be a mix. For standard game the 9700x should have a small lead due to the Zen 5 improvements. For things that are cache sensitive that will make the 7800X3D more competitive or even faster.
It makes sense to compare the 7800X3D to the 9700X due to price, and I'm sure it'll be compared against the 9800X3D when it comes out too. These aren't mutually exclusive.
Seriously? Did none of you pay attention to the charts? There's clearly a 7800X3D in all but 2 of the charts on this page alone. I could maybe see an argument made for the 7900X3D (coincidentally, the CPU I have which I why I noted it's missing) but not the 7800X3D.
Yes, I see it in the charts, hence my comment about it making sense being in here. It's similar in price to the 9700X, so it makes sense to compare from that metric. What's so difficult to understand?
I could make an argument too that an 8c/16t CPU makes more sense to compare to than the 7900X3D with 12c/24t.
This is going to be more divergent with DDR6 as the draft specs have four 16 bit sub channels for a 64 bit DIMM. However the DIMM format might only bee seen in servers with consumer products likely moving to a version CAMM which would ultimately have eight 16 bit sub channels for a 128 bit wide CAMM product.
If CAMM is ever going to take off in desktops they're going to have to come up with a vertical-oriented version (similar to how DIMMs are inserted vertically into a motherboard). There's just not enough horizontal space on ATX motherboards for multiple CAMM boards to be attached.
Would also be nice if someone came up with a vertical M.2 slot for NVMe drives.
Either that or extended ATX (or larger) motherboards are going to have to make a comeback. :)
Get off my lawn! Geez I feel old now. These are "Dual Inline Memory Modules," but otherwise just as Ryan already explained. What threw me was that they've been called dual channel for so long calling them quad channel now is misleading.
The DIMMs started back in the x486 days I think, maybe the Pentium? From 8086, 286, 386, 486, Intel increased the channel width(or the word length that the CPU can process), 8-bit, 16-bit, 32-bit, then 32-bit x2. Processors calculated smaller chunks back then, but have mostly stayed at 32-bit, which are four 8-bit bytes, so it is a 4-byte word that equals 32-bit. The Processor is much faster than the memory, so they decided to double up on the data, hence, Dual inline memory modules. Before this they were SIMMs. But I don't believe we got "dual channel" (where we got A and B channels) until 64-bit CPU's, which use twice the data, And Quad channel is mostly found in servers and HEDT systems. So, in the end this seems to be a marketing decision meant to confuse people. Sales guys don't need us to understand, just buy, and we all like MOWAR Power eh? Even if we just think it is.
This architecture has some very serious bottlenecks. It performs slightly better than Zen4 or the same in almost everything, except where AVX512 is used (DL/AI software) there the performance shoots ahead.
Disappointed to see such a wide design not deliver what it promised.
It was my first thought, but just look at Zen1-4 history. It kept the same width, but increased IPC 1.5x by going deeper (i.e. larger ROB and so on). It's the way AMD reduces their expenses - they increase width once and then slowly make CPU deeper to get small gains every year. So, I expect that Zen8 or so will be 1.5x faster that Zen4 by finally making it as wide as Apple M1.
Zen5 has 6 ALUs - 1.5x more than Zen4. e.g. Apple M1 also had 6 ALUs, but Zen5 is nowhere near M1 IPC or 1.5x Zen4 IPC. even in the official benchmarks IPC improved only by 16% on average
> It performs slightly better than Zen4 or the same in almost everything, > except where AVX512 is used
SPECint Rate-1: a 13.2% improvement is certainly respectable.
If you look at the sub-scores, 548.exchange2_r improves by a whopping 23.2%! I'm quite pleased that gcc improved by 17%.
Given that they didn't increase core count, hardly changed cache sizes, made a fairly small change in the process node, and actually reduced energy consumption (not to mention launch prices), I think it's pretty impressive!
You and so many other people have missed that because Zen5 has lower TDP ratings by default, and PBO is also disabled by default, you are looking at higher efficiency as the "out of box" experience, but with very little performance increase. If you turn on PBO and also use the additional things, such as with faster RAM, Zen5 starts to show better performance.
The Ryzen 9 9950X is the only Zen5 chip that will come with the 170W TDP seen with Zen4, and there will probably be a larger jump in performance as a result. Just wait until next week to see if I am correct.
Meh, I bet you’re referring to games because otherwise your comment makes 0 sense. And with games we need a minimum of 20 games to see the difference and not 5 or less like in here, so you basically got no point at all.
I really appreciate the limit to the power draw on Intel chips. I use systems for office work, and it is frankly ridiculous to compare a 90W CPU to a 400 W one. Apart from the power draw which is limited by the battery backup, the super high prices of the 400W desktop CPU ecosystem just does not make sense. If I need that performance I would invest in a HEDT system.
IMHO the reason why 8-core AMD cpus are compared with top Intel CPUs is that Zen5 are newer and better CPUs. So we want to see whether mid-level Zen5 can outperform top Alderlakes, despite much lower price. It's the standard practice - compare new mid-level cpus with prev-gen tops to see how far we gone.
Starting with 13th Gen Intel had a core and total thread count advantage on type vs type (Ryzen 5 vs i5 for example). However, when you compare Ryzen vs Core on thread counts the Ryzens tend to be faster for the same number of threads.
You are only partially correct. Intel has those "efficiency" cores, and includes those in the core count. E-cores are very low performance, so 16 Zen4 cores vs. 8 P-cores+16 E-cores is the comparison you are looking at, and at this point, a 7950X vs the 14900k makes for an interesting comparison in Cinebench with a generally stock configuration(turn on XMP/EXPO memory).
The PS5/ Xbox series made 8c/16t the new minimum when they launched in 2020. It's 2024 and the new gen CPU is still launching with a 6c/12t part as the "baseline". No core count increase since the original Zen launch. It's high time AMD moves to 8/16 for the Ryzen 5
Why, so 99.9% of games can ignore the 2 extra cores? Aside, the 6 core is a harvested 8 core product, it will always exist.
You cry for more cores but don’t understand the way games work today and where the development goes. Game devs don’t like high core counts and rarely optimise for more than 4-6 cores.
Actually original Zen launched (or rather, the lineup was completed shortly after launch) with 4C/4T and 4C/8T and 6C/6T. Even the original Ryzen 5 was three months after Ryzen 7 - I know because I waited for a 1600.
Zen 1 topped out at 8c/16t and you could get 16c/32t if you got a Threadripper. Since Zen 2, the top CPU is 16c/32t before going to Threadripper. Really there aren't many consumer applications that use more threads than that. Now once Zen 6 comes we might see higher core counts per CCD as that will be on a smaller process node.
Not really… 8 core still is not much faster than 8 core and 10 core would be same speed as 8 core in ost normal use cases… When the core count becomes the bottleneck, then i agree that more cores would be nice to have. But we are not in there…
I think amd screwed up with the 9700x, this should have been a 105w processor, not a 65w one. The 7700 ran at 65w and had similar clocks, while the 7700x ran at 105w with much higher base clocks.
Yep, Skatterbencher pushed to its peak with a large gains across the board.
AMD capping this processor is a sin and a shame. Why ruin a nice 8C16T part like this... Esp when your 7700X is like in spitting distance. They sabotaged it themselves.
I hope they do not do that for Zen 6 on AM5, this socket needs a good power bump from 230WPPT to at-least 270-300W give 10950X a massive lead with higher power and not cap it for BS efficiency reasons, this is a Desktop socket not a portable BGA apple machine use and throw consumable.
This is fairly good improvement watt per watt but the big thing in the testing here is that AMD is placing these chips as "X" and not the vanilla 9600 or 9700. Yes they are rated at the same wattage as non "X" counter parts from the 7000 series but the 7600X and 7700X are a hair faster and because of their higher wattage can hit those turbo values for longer. The result is more of a wash between testing of the 9600X vs. 7600X and 9700X vs. 7700X judging from other review sites today. It is an improvement but for these chip its seems AMD didn't balance power and efficiency quiet right. Case in point is the massive amount of performance left on the table if PBO is enabled with the power limits set to the same 105W values as their 7700X and 7600X counter parts. Loosening the power a bit to 85W would have been a good midstep to demonstrate an efficiency improvement alongside a more tangible performance increase.
I am still looking forward to seeing how the 9950X and 9900X fair in comparison to their 7950X and 7900X counter parts. There is additional power room at the top with the 7950X looking to get real world performance increases closer to the 16% average IPC increases AMD claims without the big asterisks of changing clock speeds or power limits impacting performance.
I'm very eager to see what the 9800X3D can do given that both the 5800X3D and 7800X3D before it reduced the clock speeds in conjunction between adding V-cache into the packaging. If the 9800X3D is able to keep the same base clocks as the 9700X but with V-cache added, it'd be a very, very nice performance increase over the 7800X3D. Similarly a 9950X3D would be a very impressive part, though I'd hope that AMD would simply put V-cache on top of both chiplets for this generation even if ithey had to reduce clocks a notch or two compared to the 9950X.
Why did AT did not do a PBO2 run on these ? Also why not compare them to a 7700X.
From what I was seeing across the board AMD's mistake was letting the 8C16T processor get a huge TDP power cap, from 105W to a mere 65W resulting in lack of IPC gains translation in IRL workloads from gaming to everything.
No idea why AMD make this stupid move. Zen 5 seems efficient but the lack of power envelope is bad. Esp when the x86 ISA always scales with power. This is a desktop socket not a cheap use and throw BGA garbage.
And now the OC part, with PBO2 this chip really excels it throws out that stupid 65W efficiency and performs like it should. Esp when we factor in Curve Shaper tool.
AT you should consider that new Curve Shaper in your next Zen 5 processor reviews like esp that 16C32T part, it will be a nice advantage for anyone who likes tinkering.
PBO is a form of overclocking. Given more time, it would have been nice to play with it as well, to see what the chip could do. But for our baseline testing, we do not run anything at overclocked settings.
"Also why not compare them to a 7700X"
We felt the 7700 was the more interesting and informative comparison, since it had the same TDP as the 9700X. This way we could get right down to business and see how the chips and architectures compared at what's essentially iso-power.
Anecdotal evidence, but in tweaking my personal 7700X, I found the sweet spot for full boost and highest all-core frequency to be around 100w PPT. I think AMD potentially missed an opportunity to do better in default benchmarks by not making the 9700X a 105w part, or at least a bit higher than 65w.
Absolutely - it appears that AMD could have avoided the bad reviews if this processor had been either given more power or marketed as "9700" without the X to match the 65W 7700.
The interesting question is why they didn't? Just a normal large company screw-up? Or have they found issues with the new process node that means they are not comfortable selling these processors in large number at higher power levels?
From a marketing perspective something has gone horribly wrong! :-)
Hopefully the 9800X and the 9950X will be able to maintain the single-thread performance advantage whilst trashing Intel in the multi-threaded benchmarks.
Gavin: There is an inconsistency from page 1 to page 3.
Page 1: "Also, it has a 65 W TDP. Still, both their predecessors, the Ryzen 7 7700X (8C/16T Zen 4) and the Ryzen 5 7600X (6C/12T Zen 4), have a higher 105 W TDP."
Page 3: "Given that both processors are nearly identical (8C/16T at 65 W TDP/88-90 W PPT), aside from the underlying core architecture"
So do they have identical TDP ratings? Or did the predecessors have a TDP rating 40w higher?
"The Zen 5 architecture doubles the L2 cache size to 1 MB per core over Zen 4..." Uh, I am pretty sure it did NOT do this. Zen 4 already was at 1MB L2 cache per core.
Gamers will buy 9800xwd or 7800x3d… not those,so these test are valid! When 9800x3d comes to market, i am sure that 7800x3d and even 5800x3d are in bigger role! Now this 7700 65w vs 9700x 65w makes a lot of sense. Ofcourse when 9700 will be released somewhere next year, it will be the direct successor of 7700, but in current cpus available, 7700 is the closest, even if it is second tier cpu and so, not directly comparable to 9700x. But it still is the closest. Based on what has been released. I expect to see 8 core 9800x with 105w to be released in the spring. Little bit more speed, much worse efficiency. So this release is a lot like 3000 series release was!
This is a pathetic power consumption page. Peak power? That's it? That's all you're showing? You don't even have the 13500 or 14500 in the chart! WHAT IS THIS!?
You guys have always had a strong AMD bias but this is ridiculous.
Bad or unimpressive results are a result of running Windows as the test operating system, Linux systems give universally better results. Don't blame AMD, blame Microsoft Windows for being unable to utilize the hardware effectively!
Finally a reviewer who tests 65W versus 65W, unlike all the others who test the new 65W CPUs against the 105W CPUs of the previous generation and then are disappointed when the new ones seemingly underperform.
That said, I would have liked to see them dig into the reasons as to why in a few benchmarks the new CPU's really do underperform compared to the 7000 series. I remember articles in the past where they really made an effort to investigate such results. Nothing like that now.
Also, I didn't see a reason for excluding the X3D CPUs from the gaming benchmarks. Weren't they made specifically for gaming? Strange to leave them out. Maybe the reason was buried in the other pages somewhere and I missed it.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
70 Comments
Back to Article
Golgatha777 - Wednesday, August 7, 2024 - link
Thanks for the review. I personally would have liked to see a 7800X3D in the benchmarks. Also, I would have omitted 720p and replaced it with 1440p gaming benchmarks. 1080p is perfectly fine for a synthetic benchmark to see which CPU is faster overall and a lot of gamers run 1440p.kpb321 - Wednesday, August 7, 2024 - link
Agreed. The 7800X3D should be very close in price to the 9700X so having both in the gaming charts would be pretty nice. Offhand I assume it will be a mix. For standard game the 9700x should have a small lead due to the Zen 5 improvements. For things that are cache sensitive that will make the 7800X3D more competitive or even faster.heffeque - Wednesday, August 7, 2024 - link
It doesn't make sense.The 7800X3D should be compared to the 9800X3D (which will come in a few months).
Trackster11230 - Wednesday, August 7, 2024 - link
It makes sense to compare the 7800X3D to the 9700X due to price, and I'm sure it'll be compared against the 9800X3D when it comes out too. These aren't mutually exclusive.Klober - Wednesday, August 7, 2024 - link
Seriously? Did none of you pay attention to the charts? There's clearly a 7800X3D in all but 2 of the charts on this page alone. I could maybe see an argument made for the 7900X3D (coincidentally, the CPU I have which I why I noted it's missing) but not the 7800X3D.Trackster11230 - Wednesday, August 7, 2024 - link
Yes, I see it in the charts, hence my comment about it making sense being in here. It's similar in price to the 9700X, so it makes sense to compare from that metric. What's so difficult to understand?I could make an argument too that an 8c/16t CPU makes more sense to compare to than the 7900X3D with 12c/24t.
boozed - Wednesday, August 7, 2024 - link
Seems odd that the gaming CPU is missing from the gaming resultsboozed - Thursday, August 8, 2024 - link
"Note: We are currently benchmarking more processors as we speak"I should read harder
Oxford Guy - Friday, August 16, 2024 - link
They said that they would re-test the Intel chips over the weekend in May. Did those updated benchmarks ever get done and published?frshi - Wednesday, August 7, 2024 - link
The table says Quad Channel DDR support, I don't think that's true. It's 4 slots but it's dual channel, isn't it?Ryan Smith - Wednesday, August 7, 2024 - link
This is an area where technical specifications and casual nomenclature have drifted apart.DDR5 channels are 32-bits each. A DIMM offers two 32-bit channels, for 64-bits altogether.
So AM5 takes two DIMMs. But it's technically four independent DDR5 channels.
Kevin G - Wednesday, August 7, 2024 - link
This is going to be more divergent with DDR6 as the draft specs have four 16 bit sub channels for a 64 bit DIMM. However the DIMM format might only bee seen in servers with consumer products likely moving to a version CAMM which would ultimately have eight 16 bit sub channels for a 128 bit wide CAMM product.phoenix_rizzen - Tuesday, August 13, 2024 - link
If CAMM is ever going to take off in desktops they're going to have to come up with a vertical-oriented version (similar to how DIMMs are inserted vertically into a motherboard). There's just not enough horizontal space on ATX motherboards for multiple CAMM boards to be attached.Would also be nice if someone came up with a vertical M.2 slot for NVMe drives.
Either that or extended ATX (or larger) motherboards are going to have to make a comeback. :)
'nar - Friday, August 16, 2024 - link
Get off my lawn! Geez I feel old now. These are "Dual Inline Memory Modules," but otherwise just as Ryan already explained. What threw me was that they've been called dual channel for so long calling them quad channel now is misleading.The DIMMs started back in the x486 days I think, maybe the Pentium? From 8086, 286, 386, 486, Intel increased the channel width(or the word length that the CPU can process), 8-bit, 16-bit, 32-bit, then 32-bit x2. Processors calculated smaller chunks back then, but have mostly stayed at 32-bit, which are four 8-bit bytes, so it is a 4-byte word that equals 32-bit. The Processor is much faster than the memory, so they decided to double up on the data, hence, Dual inline memory modules. Before this they were SIMMs. But I don't believe we got "dual channel" (where we got A and B channels) until 64-bit CPU's, which use twice the data, And Quad channel is mostly found in servers and HEDT systems. So, in the end this seems to be a marketing decision meant to confuse people. Sales guys don't need us to understand, just buy, and we all like MOWAR Power eh? Even if we just think it is.
Terry_Craig - Wednesday, August 7, 2024 - link
This architecture has some very serious bottlenecks. It performs slightly better than Zen4 or the same in almost everything, except where AVX512 is used (DL/AI software) there the performance shoots ahead.Disappointed to see such a wide design not deliver what it promised.
yeeeeman - Wednesday, August 7, 2024 - link
yes, exactlyBulat Ziganshin - Wednesday, August 7, 2024 - link
It was my first thought, but just look at Zen1-4 history. It kept the same width, but increased IPC 1.5x by going deeper (i.e. larger ROB and so on). It's the way AMD reduces their expenses - they increase width once and then slowly make CPU deeper to get small gains every year. So, I expect that Zen8 or so will be 1.5x faster that Zen4 by finally making it as wide as Apple M1.Bulat Ziganshin - Wednesday, August 7, 2024 - link
sorry, I meant "Zen8 will be 1.5x faster than Zen4 by making it as DEEP as M1 while keeping the same width as Zen5"Khanan - Wednesday, August 7, 2024 - link
Did we read the same article? Maybe you shouldn’t comment if you didn’t read or understand the article.And if you’re only about games, 5 games are nothing, go for the reviews where 20 games are tested (at least).
Bulat Ziganshin - Wednesday, August 7, 2024 - link
Zen5 has 6 ALUs - 1.5x more than Zen4. e.g. Apple M1 also had 6 ALUs, but Zen5 is nowhere near M1 IPC or 1.5x Zen4 IPC. even in the official benchmarks IPC improved only by 16% on averageKhanan - Wednesday, August 7, 2024 - link
M1 is irrelevant as Apple is more or less completely irrelevant to PCs and 100% to servers. And I didn’t talk to you to begin with.schujj07 - Thursday, August 8, 2024 - link
You're right Zen 5 is no where near the M1 in IPC it is actually near M3.mode_13h - Thursday, August 8, 2024 - link
> It performs slightly better than Zen4 or the same in almost everything,> except where AVX512 is used
SPECint Rate-1: a 13.2% improvement is certainly respectable.
If you look at the sub-scores, 548.exchange2_r improves by a whopping 23.2%! I'm quite pleased that gcc improved by 17%.
Given that they didn't increase core count, hardly changed cache sizes, made a fairly small change in the process node, and actually reduced energy consumption (not to mention launch prices), I think it's pretty impressive!
Targon - Friday, August 9, 2024 - link
You and so many other people have missed that because Zen5 has lower TDP ratings by default, and PBO is also disabled by default, you are looking at higher efficiency as the "out of box" experience, but with very little performance increase. If you turn on PBO and also use the additional things, such as with faster RAM, Zen5 starts to show better performance.The Ryzen 9 9950X is the only Zen5 chip that will come with the 170W TDP seen with Zen4, and there will probably be a larger jump in performance as a result. Just wait until next week to see if I am correct.
Khanan - Saturday, August 10, 2024 - link
Meh, I bet you’re referring to games because otherwise your comment makes 0 sense. And with games we need a minimum of 20 games to see the difference and not 5 or less like in here, so you basically got no point at all.Bulat Ziganshin - Wednesday, August 14, 2024 - link
While on the paper 7950x had 170W TDP, really it was closer to 230W, so Zen5 reduced real TDP for the entire linedrajitshnew - Wednesday, August 7, 2024 - link
I really appreciate the limit to the power draw on Intel chips. I use systems for office work, and it is frankly ridiculous to compare a 90W CPU to a 400 W one.Apart from the power draw which is limited by the battery backup, the super high prices of the 400W desktop CPU ecosystem just does not make sense.
If I need that performance I would invest in a HEDT system.
Jorgp2 - Wednesday, August 7, 2024 - link
I don't think you understand what power is.You're comparing the base power of one CPU, to the turbo power of another CPU.
Khanan - Wednesday, August 7, 2024 - link
No, he’s comparing turbo to turbo power, what’s your problem? The 9700X is ~90W peak, Intels was 400W peak, 290W with the new settings.Bulat Ziganshin - Wednesday, August 7, 2024 - link
IMHO the reason why 8-core AMD cpus are compared with top Intel CPUs is that Zen5 are newer and better CPUs. So we want to see whether mid-level Zen5 can outperform top Alderlakes, despite much lower price. It's the standard practice - compare new mid-level cpus with prev-gen tops to see how far we gone.schujj07 - Thursday, August 8, 2024 - link
Starting with 13th Gen Intel had a core and total thread count advantage on type vs type (Ryzen 5 vs i5 for example). However, when you compare Ryzen vs Core on thread counts the Ryzens tend to be faster for the same number of threads.Targon - Friday, August 9, 2024 - link
You are only partially correct. Intel has those "efficiency" cores, and includes those in the core count. E-cores are very low performance, so 16 Zen4 cores vs. 8 P-cores+16 E-cores is the comparison you are looking at, and at this point, a 7950X vs the 14900k makes for an interesting comparison in Cinebench with a generally stock configuration(turn on XMP/EXPO memory).meacupla - Wednesday, August 7, 2024 - link
is there an idle power draw graph?Slash3 - Wednesday, August 7, 2024 - link
4K gaming page still has a few rather unnecessary text padding entries at the bottom, FYI. :)isthisavailable - Wednesday, August 7, 2024 - link
The PS5/ Xbox series made 8c/16t the new minimum when they launched in 2020. It's 2024 and the new gen CPU is still launching with a 6c/12t part as the "baseline". No core count increase since the original Zen launch. It's high time AMD moves to 8/16 for the Ryzen 5Khanan - Wednesday, August 7, 2024 - link
Why, so 99.9% of games can ignore the 2 extra cores? Aside, the 6 core is a harvested 8 core product, it will always exist.You cry for more cores but don’t understand the way games work today and where the development goes. Game devs don’t like high core counts and rarely optimise for more than 4-6 cores.
erotomania - Thursday, August 8, 2024 - link
Actually original Zen launched (or rather, the lineup was completed shortly after launch) with 4C/4T and 4C/8T and 6C/6T. Even the original Ryzen 5 was three months after Ryzen 7 - I know because I waited for a 1600.schujj07 - Friday, August 9, 2024 - link
Zen 1 topped out at 8c/16t and you could get 16c/32t if you got a Threadripper. Since Zen 2, the top CPU is 16c/32t before going to Threadripper. Really there aren't many consumer applications that use more threads than that. Now once Zen 6 comes we might see higher core counts per CCD as that will be on a smaller process node.haukionkannel - Saturday, August 10, 2024 - link
Not really… 8 core still is not much faster than 8 core and 10 core would be same speed as 8 core in ost normal use cases…When the core count becomes the bottleneck, then i agree that more cores would be nice to have. But we are not in there…
shabby - Wednesday, August 7, 2024 - link
I think amd screwed up with the 9700x, this should have been a 105w processor, not a 65w one. The 7700 ran at 65w and had similar clocks, while the 7700x ran at 105w with much higher base clocks.Khanan - Wednesday, August 7, 2024 - link
If you are right (big if, you’re probably not), this just makes it a great overclocker via PBO.shabby - Wednesday, August 7, 2024 - link
It does, der8auer hit like 170w on his 9700x.https://www.youtube.com/watch?v=jPJ0Khw3kIc
Silver5urfer - Thursday, August 8, 2024 - link
Yep, Skatterbencher pushed to its peak with a large gains across the board.AMD capping this processor is a sin and a shame. Why ruin a nice 8C16T part like this... Esp when your 7700X is like in spitting distance. They sabotaged it themselves.
I hope they do not do that for Zen 6 on AM5, this socket needs a good power bump from 230WPPT to at-least 270-300W give 10950X a massive lead with higher power and not cap it for BS efficiency reasons, this is a Desktop socket not a portable BGA apple machine use and throw consumable.
Khanan - Thursday, August 8, 2024 - link
if that's true they can fix it with a 7700XT (like in 2nd gen).Khanan - Thursday, August 8, 2024 - link
*9700XTschujj07 - Friday, August 9, 2024 - link
The more power you use the harder it is to cool. Efficiency is also very important.Oxford Guy - Friday, August 16, 2024 - link
AMD could also simply by delivering the minimal added value it thinks it can. Coasting to profit.Kevin G - Wednesday, August 7, 2024 - link
This is fairly good improvement watt per watt but the big thing in the testing here is that AMD is placing these chips as "X" and not the vanilla 9600 or 9700. Yes they are rated at the same wattage as non "X" counter parts from the 7000 series but the 7600X and 7700X are a hair faster and because of their higher wattage can hit those turbo values for longer. The result is more of a wash between testing of the 9600X vs. 7600X and 9700X vs. 7700X judging from other review sites today. It is an improvement but for these chip its seems AMD didn't balance power and efficiency quiet right. Case in point is the massive amount of performance left on the table if PBO is enabled with the power limits set to the same 105W values as their 7700X and 7600X counter parts. Loosening the power a bit to 85W would have been a good midstep to demonstrate an efficiency improvement alongside a more tangible performance increase.I am still looking forward to seeing how the 9950X and 9900X fair in comparison to their 7950X and 7900X counter parts. There is additional power room at the top with the 7950X looking to get real world performance increases closer to the 16% average IPC increases AMD claims without the big asterisks of changing clock speeds or power limits impacting performance.
I'm very eager to see what the 9800X3D can do given that both the 5800X3D and 7800X3D before it reduced the clock speeds in conjunction between adding V-cache into the packaging. If the 9800X3D is able to keep the same base clocks as the 9700X but with V-cache added, it'd be a very, very nice performance increase over the 7800X3D. Similarly a 9950X3D would be a very impressive part, though I'd hope that AMD would simply put V-cache on top of both chiplets for this generation even if ithey had to reduce clocks a notch or two compared to the 9950X.
HideOut - Wednesday, August 7, 2024 - link
"AMD has also taken a bit of a different approach with AVX-512 instructions for Zen 4,"You mean zen 5
mukiex - Thursday, August 8, 2024 - link
"Zen 5 is alive"No disassemble Zen 5!
LarsBars - Thursday, August 8, 2024 - link
Are you still planning on including the core-to-core latency testing?Silver5urfer - Thursday, August 8, 2024 - link
I second that, it would be great to see a proper deep dive into Zen 5. Please do that with upcoming bigger core parts.Silver5urfer - Thursday, August 8, 2024 - link
Why did AT did not do a PBO2 run on these ? Also why not compare them to a 7700X.From what I was seeing across the board AMD's mistake was letting the 8C16T processor get a huge TDP power cap, from 105W to a mere 65W resulting in lack of IPC gains translation in IRL workloads from gaming to everything.
No idea why AMD make this stupid move. Zen 5 seems efficient but the lack of power envelope is bad. Esp when the x86 ISA always scales with power. This is a desktop socket not a cheap use and throw BGA garbage.
And now the OC part, with PBO2 this chip really excels it throws out that stupid 65W efficiency and performs like it should. Esp when we factor in Curve Shaper tool.
AT you should consider that new Curve Shaper in your next Zen 5 processor reviews like esp that 16C32T part, it will be a nice advantage for anyone who likes tinkering.
Ryan Smith - Friday, August 9, 2024 - link
"Why did AT did not do a PBO2 run on these ?"PBO is a form of overclocking. Given more time, it would have been nice to play with it as well, to see what the chip could do. But for our baseline testing, we do not run anything at overclocked settings.
"Also why not compare them to a 7700X"
We felt the 7700 was the more interesting and informative comparison, since it had the same TDP as the 9700X. This way we could get right down to business and see how the chips and architectures compared at what's essentially iso-power.
Golgatha777 - Friday, August 9, 2024 - link
Anecdotal evidence, but in tweaking my personal 7700X, I found the sweet spot for full boost and highest all-core frequency to be around 100w PPT. I think AMD potentially missed an opportunity to do better in default benchmarks by not making the 9700X a 105w part, or at least a bit higher than 65w.OFelix - Saturday, August 10, 2024 - link
Absolutely - it appears that AMD could have avoided the bad reviews if this processor had been either given more power or marketed as "9700" without the X to match the 65W 7700.The interesting question is why they didn't? Just a normal large company screw-up? Or have they found issues with the new process node that means they are not comfortable selling these processors in large number at higher power levels?
From a marketing perspective something has gone horribly wrong! :-)
Hopefully the 9800X and the 9950X will be able to maintain the single-thread performance advantage whilst trashing Intel in the multi-threaded benchmarks.
Targon - Friday, August 9, 2024 - link
You could say that XMP is a form of overclocking, but no one has a problem turning THAT on when benchmarking.Zoolook13 - Wednesday, August 14, 2024 - link
Except for Anandtech, they are consistent.James5mith - Friday, August 9, 2024 - link
Gavin: There is an inconsistency from page 1 to page 3.Page 1: "Also, it has a 65 W TDP. Still, both their predecessors, the Ryzen 7 7700X (8C/16T Zen 4) and the Ryzen 5 7600X (6C/12T Zen 4), have a higher 105 W TDP."
Page 3: "Given that both processors are nearly identical (8C/16T at 65 W TDP/88-90 W PPT), aside from the underlying core architecture"
So do they have identical TDP ratings? Or did the predecessors have a TDP rating 40w higher?
Ryan Smith - Friday, August 9, 2024 - link
The comment on page 1 is in reference to the 7700X. That is a 105W TDP processor.The comment on page 3 is in reference to the vanilla (non-X) 7700. That is a 65W TDP processor.
NextGen_Gamer - Friday, August 9, 2024 - link
"The Zen 5 architecture doubles the L2 cache size to 1 MB per core over Zen 4..." Uh, I am pretty sure it did NOT do this. Zen 4 already was at 1MB L2 cache per core.nightbird321 - Saturday, August 10, 2024 - link
Gaming benchmarks without a single X3D in the comparison, how AT has fallen.haukionkannel - Saturday, August 10, 2024 - link
Gamers will buy 9800xwd or 7800x3d… not those,so these test are valid!When 9800x3d comes to market, i am sure that 7800x3d and even 5800x3d are in bigger role!
Now this 7700 65w vs 9700x 65w makes a lot of sense. Ofcourse when 9700 will be released somewhere next year, it will be the direct successor of 7700, but in current cpus available, 7700 is the closest, even if it is second tier cpu and so, not directly comparable to 9700x. But it still is the closest. Based on what has been released. I expect to see 8 core 9800x with 105w to be released in the spring. Little bit more speed, much worse efficiency. So this release is a lot like 3000 series release was!
nightbird321 - Sunday, August 11, 2024 - link
These benchmarks are meant to make Intel to look to be the best CPUs for gaming, which is a laughable conclusion.Hrel - Tuesday, August 13, 2024 - link
This is a pathetic power consumption page. Peak power? That's it? That's all you're showing? You don't even have the 13500 or 14500 in the chart! WHAT IS THIS!?You guys have always had a strong AMD bias but this is ridiculous.
Oxford Guy - Friday, August 16, 2024 - link
'You guys have always had a strong AMD bias'Poppycock.
Heavensrevenge - Thursday, August 15, 2024 - link
Bad or unimpressive results are a result of running Windows as the test operating system, Linux systems give universally better results.Don't blame AMD, blame Microsoft Windows for being unable to utilize the hardware effectively!
https://www.phoronix.com/review/amd-ryzen-9950x-99... for a REAL review.
PProchnow - Friday, August 16, 2024 - link
-----2699 single -----> 7600--------11588 multi ---> 7600
------- 3228 single-----> 9600
-------13379 multi ------> 9600
Sivar - Tuesday, August 20, 2024 - link
"Zen 5 is alive!"I see what you did there, Johnny 5.
Bonez0r - Tuesday, August 27, 2024 - link
Finally a reviewer who tests 65W versus 65W, unlike all the others who test the new 65W CPUs against the 105W CPUs of the previous generation and then are disappointed when the new ones seemingly underperform.That said, I would have liked to see them dig into the reasons as to why in a few benchmarks the new CPU's really do underperform compared to the 7000 series. I remember articles in the past where they really made an effort to investigate such results. Nothing like that now.
Also, I didn't see a reason for excluding the X3D CPUs from the gaming benchmarks. Weren't they made specifically for gaming? Strange to leave them out. Maybe the reason was buried in the other pages somewhere and I missed it.