News Ryzen 7 5800X3D Beats Ryzen 7 5800X By 9% In Geekbench 5

For high performance gaming though, single core performance tends to be where it's at, and seeing a weaker show here doesn't lead me to believe it will be as good as AMD claims, except maybe in well implemented DX12/Vulkan games running on a Radeon GPU.
 
May 21, 2021
126
66
170
So why does the cache help in the multicore test but not the single core one?
My guess would be the ability of the L3 cache to hold more data that all the cores need. With just single core only the current jobs data is stored on the cache, when it’s done with that job the cache is refreshed with the new single core job and so on and so on. Since the multi core test will run a different job on each core simultaneously, having more L3 allows more of the individual data each core needs for its respective job to be stored in cache which means less cache misses being sent to RAM access which is about 6-7x higher latency vs cache access. TLDR, a big enough L3 allows all cores to be fed data at low latency on the 5800x3d vs resorting to high latency ram when the 5800x’s L3 cache is full.
 
  • Like
Reactions: tennis2
So why does the cache help in the multicore test but not the single core one?
We can look back at AnandTech's deep dive on Zen 3 to get answers for this.
  • It's important to note that the L3 cache is a victim cache of L2. Basically, when L2 gets full, the oldest thing gets kicked out into L3. Also as a result, L3 cannot prefetch data.
  • Zen 3's increased cache came at a cost of higher latency and lower bandwidth per core. It was meant more to service multi-core workloads, since this allows every core in the CCX to have the same latency when accessing each other's data
  • They made the observation that the L2 TLB is only 2K entries long, with each entry covering 4KiB. This means per core, the TLB can only cover 8MiB of L3 cache. So a single core has issues even without the extra cache from V-Cache
There's also another issue: you can only have so much cache before you'll spend too much time looking through cache than it would've taken to look it up from RAM.

For example if we look at this chart from AnandTech's article:
lat-5950.png

(note the y-axis is logarithmic)

If we assume the latency is linear after 16MiB, which appears to be 15ns in the worst case, and L3 cache gets full, you're now looking at 90ns just to look through the whole thing. It would've been faster to go look at RAM at that point.
 
Last edited:
May 21, 2021
126
66
170
We can look back at AnandTech's deep dive on Zen 3 to get answers for this.
  • It's important to note that the L3 cache is a victim cache of L2. Basically, when L2 gets full, the oldest thing gets kicked out into L3. Also as a result, L3 cannot prefetch data.
  • Zen 3's increased cache came at a cost of higher latency and lower bandwidth per core. It was meant more to service multi-core workloads, since this allows every core in the CCX to have the same latency when accessing each other's data
  • They made the observation that the L2 TLB is only 2K entries long, with each entry covering 4KiB. This means per core, the TLB can only cover 8MiB of L3 cache. So a single core has issues even without the extra cache from V-Cache
There's also another issue: you can only have so much cache before you'll spend too much time looking through cache than it would've taken to look it up from RAM.

For example if we look at this chart from AnandTech's article:
lat-5950.png

(note the y-axis is logarithmic)

If we assume the latency is linear after 16MiB, which appears to be 15ns in the worst case, and L3 cache gets full, you're now looking at 90ns just to look through the whole thing. It would've been faster to go look at RAM at that point.
Yes you make valid points and you are certainly learned in the sci-art of micro architecture, however I would like to correct a few things you bring up that might help you understand why bigger L3 is helpful in certain cpu workloads. First, the ram latencies you are referencing (less than 90ns) are the best case scenario IE data is requested on an open active row and column which is prepped to be read. This is almost never the case. Aida 64 and other latency and bandwidth benchmarks always give the best case latency for cache and ram so you must take each to be equivalent in how their respective latencies react to different data locations IE my ryzen 9 5950x L3 has a 10.4ns latency according to Aida 64 but as the graph you provided shows worst case is closer to 90ns. The same goes for RAM, my cas 14 3800mhz ram has an Aida 64 latency of 60.2ns but worst case is closer to 1 us or more due to row precharge, access, read, etc. duties to find and transfer the data + a trfc refresh cycle that can halt the data fetch in its tracks and cause the fetch to restart.

Second, a victim cache (unlike inclusive cache) is simply a term used to describe an overflow cache. You are correct that the oldest data in L1/2 gets booted to L3 but second, just as important, is if the jobs data is too big for L1/2 (which often is the case) the L3 acts as a slower L2 cache extension to allow more of the needed data for the job at hand from having to be continually fetched from the much slower ram then eliminated from cache from disuse only for the deleted data a microsecond or so later being re-fetched from RAM again due to many workloads re-using the same data multiple times (obviously like you said, if said data value is popular it will stay in L2 but if it’s disused for more than a couple microseconds then it is booted into L3 and eventually deleted as L2 continues to push disused data into L3, thus having more L3 prolongs the disused data life in cache in case it’s called upon again, which happens more often than you think). Thus, as I surmised in my answer, having more L3 cache helps with multi core workloads by making ram access a rarer instance while feeding the cores with more consistent data flow. That is why the single core scores are very similar between the 2 ryzen 5800 series cpus (because the data for 1 cores job is small enough to fit in the 5800x’s L1/2/3 but not big enough to fit all the data needed to drive 16 independent jobs at the same time. The 5800x3d has enough L3 to fit more of geekbench’s 16 thread workload and thus limits the amount of ram accesses. That is why AMD claims that the v-cache improves gaming performance while suffering from lower clock frequency. If a game thread can be kept fed with low latency cache accesses instead of high latency ram accesses then the cpu core wastes less clock cycles waiting for the required data and thus gets more done on average per clock cycle.

Any way, I hope you don’t take my rebuttal in the wrong light, you do make valid points but microarchitectures are very complex and thus their behaviors are complex as well. Unless you study architectures for a living it’s harder to gain detailed knowledge.
 

watzupken

Reputable
Mar 16, 2020
1,007
505
6,070
Actually the results are not surprising. In fact, I feel AMD should not even bother about releasing a stop gap solution here because in my opinion, this offers too little performance uplift, and too much of a price increase for most users. While it is true that Intel have regained their performance crown, but the most important thing is price. Zen 3 may be older, but if the price is right, it is still a viable option. But for USD 449, which is basically the launch price of the R7 5800X, the performance uplift is uninteresting/ unexciting, so I doubt they will see any improvement in sales, nor will they be able to retain the performance crown.
 
  • Like
Reactions: JamesJones44

salgado18

Distinguished
Feb 12, 2007
928
373
19,370
Actually the results are not surprising. In fact, I feel AMD should not even bother about releasing a stop gap solution here because in my opinion, this offers too little performance uplift, and too much of a price increase for most users. While it is true that Intel have regained their performance crown, but the most important thing is price. Zen 3 may be older, but if the price is right, it is still a viable option. But for USD 449, which is basically the launch price of the R7 5800X, the performance uplift is uninteresting/ unexciting, so I doubt they will see any improvement in sales, nor will they be able to retain the performance crown.
I don't think it is a stop gap, but more of a prototype test. If this works well in real world use and with consumers, the tech can be used in future processors, with lowered prices and better performance. If it doesn't, this will be the first and last 3D V-Cache CPU.
 
  • Like
Reactions: tennis2

JamesJones44

Reputable
Jan 22, 2021
620
560
5,760
Probably should have just cancelled this thing after the initial engineering samples didn't pan out. I doesn't look like from these early benchmarks this chip fits in anywhere, especially since it can't be overclocked at all.
 
I don't think it is a stop gap, but more of a prototype test. If this works well in real world use and with consumers, the tech can be used in future processors, with lowered prices and better performance. If it doesn't, this will be the first and last 3D V-Cache CPU.
The problem I'm seeing is AMD hyped this up to be some sort of game changer and it'll bring it appreciable performance upgrades over the non V-cache CPUs. While I don't disagree they wanted to throw this tech out on something to see how it works, I think they should've released it with less fanfare.
 
Graphics Command Centre.
The problem I'm seeing is AMD hyped this up to be some sort of game changer and it'll bring it appreciable performance upgrades over the non V-cache CPUs. While I don't disagree they wanted to throw this tech out on something to see how it works, I think they should've released it with less fanfare.

If you can get 9-15% IPC gain without an architectural change, then it's a win for consumers. If they put 3D cache on future chips and try to solve the lower clock speed issues. Then it's a winning feature. It's at the point where it's showing great promise, but it's not a game changer unless the cpu clock can match current offerings.
 
I think they should've released it with less fanfare.
You don't sell product by timid marketing. Unfortunately, yes, lots of hype can create unreal expectations/excitement amongst consumers. One thing is for certain, AMD/Intel/Nvidia are going to show cherry-picked metrics in their marketing slides that paint the respective product in the best light.

Also, its a fairly novel design/fab achievement (even though AMD is just operating within the capabilities of TSMC and moreover the equipment that TSMC uses). They deserve to be proud. This is also a bit of a risk for AMD. I can't imagine this comes as a no-cost alternative to the 5800X. It costs more to fab and will therefore command a price premium regardless of performance. AMD has/had to make careful decisions regarding whether the perceived performance uplift is justifiable by the > cost and can support favorable profit margins. Keep in mind, they'll have a break-even volume for the 5800X3D as well. If it tanks, it hurts.

As discussed above, Vcache can't/shouldn't be a silver bullet to all aspects of CPU performance. Like we see on the single-core geekbench score, there will be scenarios where little/no gains are realized. That just means that the higher level caches are sufficiently proportioned/designed to handle said workloads. Don't get me wrong, you still have to get benchmark "wins", but the important part is realizing what metrics are most indicative of "real world" workloads for some/most users, and what metrics are more niche.
 
Last edited:
You don't sell product by timid marketing. Unfortunately, yes, lots of hype can create unreal expectations/excitement amongst consumers. One thing is for certain, AMD/Intel/Nvidia are going to show cherry-picked metrics in their marketing slides that paint the respective product in the best light.

Also, its a fairly novel design/fab achievement. They deserve to be proud.

As discussed above, Vcache can't/shouldn't be a silver bullet to all aspects of CPU performance. Like we see on the single-core geekbench score, there will be scenarios where little/no gains are realized. That just means that the higher level caches are sufficiently proportioned/designed to handle said workloads. Don't get me wrong, you still have to get benchmark "wins", but the important part is realizing what metrics are most indicative of "real world" workloads for some/most users, and what metrics are more niche.
The problem I have is AMD seems to be positing the 5800X3D has a gaming CPU that will take back the performance crown (or if not, at least stand on the podium). Most games, especially with esports titles, can't take full advantage of an 8C/16T processor yet. So while sure, the 5800X3D may be able to beat out a 5800X in multicore loads, that won't translate much to games. And depending on the game or API used, single core performance is still going to be the deciding factor in how many frames per second you get. And then we have the apparent snafu that AMD teased a 5900/5950X with 3D V-Cache, but decided to release only a 5800X.

And I really don't get who this product is for anymore. As the article points out, if you want higher multicore performance, the 5900X can be had for around the same price. I'm convinced AMD wanted a beta test V-cache and the only way to convince people to buy it was to hype it up.

If anything, they should've fed this to system builders.
 
It may not be everything it was said to be, but up to 10% more performance at lower clock speeds is pretty great.
This is where I am at. Coming from the 3700X the jump to a 5800X3D if it averages a 10% increase over the 5800X/5900X that would be a worthwhile improvement for me running a 240Hz monitor and no need to buy a new motherboard. It all hangs on the reviews.
 

JamesJones44

Reputable
Jan 22, 2021
620
560
5,760
This is where I am at. Coming from the 3700X the jump to a 5800X3D if it averages a 10% increase over the 5800X/5900X that would be a worthwhile improvement for me running a 240Hz monitor and no need to buy a new motherboard. It all hangs on the reviews.

It's really only 10% above the 5800x though. I'm not sure why you would drop an extra $100 over the 5800x for something that gaming wise is equal, but inferior multithreading wise (5800X3D vs 5900x). Drop the $100 on the 5900x and you've got a better CPU than the 5800X3D.
 
  • Like
Reactions: spentshells

Math Geek

Titan
Ambassador
It's really only 10% above the 5800x though.

it's all on perspective. if this was a new intel cpu that was 10% over the last gen AMD chip, the headline would be "INTEL RETAKES THE CROWN WITH A MASSIVE BEATDOWN OF AMD!!!"

or even if it was a 10% gain from one intel gen to the next "refresh". it would still be a great fanfare as a huge increase and "just go buy it" recommendation.......
 
It's really only 10% above the 5800x though. I'm not sure why you would drop an extra $100 over the 5800x for something that gaming wise is equal, but inferior multithreading wise (5800X3D vs 5900x). Drop the $100 on the 5900x and you've got a better CPU than the 5800X3D.
$100 in the grand scheme of my rig isn’t that much, I’ve spent more on fans and lighting so a 10% lift in fps for $100 is quite appealing. If it’s genuinely as good or better than any other cpu for gaming it will give me the best performance from my AM4 motherboard, a 3700X is showing its limitation paired with a 240Hz monitor. I can then put off my next big upgrade for longer until DDR5 has matured and performance gains improved with cost dropped.
 

JamesJones44

Reputable
Jan 22, 2021
620
560
5,760
it's all on perspective. if this was a new intel cpu that was 10% over the last gen AMD chip, the headline would be "INTEL RETAKES THE CROWN WITH A MASSIVE BEATDOWN OF AMD!!!"

or even if it was a 10% gain from one intel gen to the next "refresh". it would still be a great fanfare as a huge increase and "just go buy it" recommendation.......

I agree, but that's not what I'm saying. I'm saying if you were going to drop an extra $100 on a CPU, why not pick the 5900x which gives you just about the same 10% jump for games plus gives you a CPU that can be over clocked and provides better multi-thread performance? To me the value proposition for the 5800X3D doesn't make sense given its price.