• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

Info 64MB V-Cache on 5XXX Zen3 Average +15% in Games

Page 20 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
K

Kedas

Senior member
Well we know now how they will bridge the long wait to Zen4 on AM5 Q4 2022.
Production start for V-cache is end this year so too early for Zen4 so this is certainly coming to AM4.
+15% Lisa said is "like an entire architectural generation"
 
Last edited:
jpiniero said:
They might do a new stepping but you also have to fab the cache die. I have no idea how long the cache die would take. Then you have the bonding process.
Click to expand...

This looks like a first sighting of the B2 stepping today. If AMD using only this stepping for turning Zen 3 into Zen 3D, then yeah, it will need the full fab cycle. And we don't know when that started (or will start).

The first mentions of the new stepping was ~ May. So it may have already entered production.

As far as SRAM die, fabrication of that would be in parallel, not sequential to base die, and would take shorter than base die, so not a bottleneck.

Then the only additional time with the assembly of stacked dies.

https://twitter.com/x/status/1430962765846523911
 
The 3d cache versions of Zen 3 CCDs have to be thinned, which they're not going to do on a per die basis so anything already fabricated and diced won't be eligible to be used. Most likely the fab is doing the thinning so any wafers already sent to AMD are also probably not eligible. That's not to say AMD couldn't already have produced thinned wafers in preparation, but any 'standard flow' dice won't be used. The cache layer fab time should be more or less the same as the CCDs.
 
Hitman928 said:
The 3d cache versions of Zen 3 CCDs have to be thinned, which they're not going to do on a per die basis so anything already fabricated and diced won't be eligible to be used. Most likely the fab is doing the thinning so any wafers already sent to AMD are also probably not eligible. That's not to say AMD couldn't already have produced thinned wafers in preparation, but any 'standard flow' dice won't be used. The cache layer fab time should be more or less the same as the CCDs.
Click to expand...

Good point about thinning most likely taking place at wafer stage. But still, that's likely one of the last stages of the wafer processing - whether to send a batch of wafers for thinning or not.

As far as wafers sent to AMD - anything that has left TSMC without V-Cache is obviously not going to have V-Cache, since the bonding is done at TSMC facility.
 
eek2121 said:

Intel Corporation Alder Lake Client Platform - Geekbench Browser

Benchmark results for an Intel Corporation Alder Lake Client Platform with an Intel Core i9-12900K processor.
browser.geekbench.com

1893/17299
That is the 12900k. Maybe AMD will drop some more definite info soon. Competition is grand.
Click to expand...

Does not look that impressive vs a properly tuned 5950X system. Those numbers they pull from the geekbench site are much lower than what I see on my system also.

Page 510 - Discussion - Intel current and future Lakes & Rapids thread

Page 510 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
forums.anandtech.com forums.anandtech.com
 
jamescox said:
People have stopped bothering to argue with you on this since you are making up straw-man arguments. In your mind, AMD is giving up on single thread performance if they don’t or can’t release a 288 MB L3 cache variant for the consumer market? No one said or implied that other than you. I personally think they will do fine with 96 MB. A larger cache variant Ryzen part would break market segmentation and doesn’t have any place in the stack. What would it be priced at vs. threadripper 5000 parts that could have high cache per core size without X3D?

Anyway, it likely isn’t relevant since the more than 1 layer device probably isn’t going to be available in the same time frame. I know people trying to get Milan processors at the moment and they can’t get them. This is very bad in the server market when a company specs a machine, gets evaluation machines, and then can’t actually get the parts to build the production machines. Perhaps covid is to blame for the delays; I don’t know. Threadripper 5000 isn’t going to be available until AMD can mostly satisfy Milan demand. Current rumors say threadripper 5000 availability will not be until November now. Milan-X3D (presumably 4+ layer devices) is rumored to not be available until sometime in 2022.

So, I guess from what you seem to believe, AMD is just completely SOL, since they are probably only going to have 96 MB L3 parts in the consumer market at the end of the year.

It has happened frequently that people in forums try to build up unreasonable expectations for upcoming AMD parts. I don’t know if this is just fanboyism or an attempt to make truly incredible AMD releases seem more like a disappointment. I am wondering which category you fall into.
Click to expand...
Thank you for still enduring this all.
 
Kedas said:
Well they will be at least be able to do a paper launch at that time, just to take all energy out of inlets launch, we all know the one that wins the gaming benchmarks gets the crown.
Click to expand...

We don't know if it will be a paper launch or not, and after hearing that only a few versions of Alder Lake will be launched, too, my opinion is that both these new CPUs will be quite limited in availability until 2022.
 
Given the processing required, I'd hope it's more than just a cache slap on. Maybe some minor improvements elsewhere not unlike Zen+, but just as a new stepping. I've seen some of the Alderlake leaks and I'm not too impressed so far. As they say, actual benches matter.
 
DrMrLordX said:
In fairness to @lobz , it certainly looks like AMD has bumped production of Zen3d forward unexpectedly. I wonder what other optimizations have been made to the B2 stepping?
Click to expand...
In fairness to me, nobody here knows if production was bumped or not, nor do they know exactly when will actual production start or not. Joe said AMD will have chips ready to counter ADL, just because in his opinion they should... After seeing him ravage a couple of threads for weeks now, I decided to wait and see.
 
lobz said:
In fairness to me, nobody here knows if production was bumped or not, nor do they know exactly when will actual production start or not. Joe said AMD will have chips ready to counter ADL, just because in his opinion they should... After seeing him ravage a couple of threads for weeks now, I decided to wait and see.
Click to expand...

Well, that Greymon55 certainly seems confident that Zen3d could be out this year, which was not what was originally teased to the public.
 
Makaveli said:
Does not look that impressive vs a properly tuned 5950X system. Those numbers they pull from the geekbench site are much lower than what I see on my system also.

Page 510 - Discussion - Intel current and future Lakes & Rapids thread

Page 510 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
forums.anandtech.com forums.anandtech.com
Click to expand...

I don’t think you realize the full implications of ADL-S. The 12900k is a $500-$600 chip, not an $800 chip. It’s competition is the 5900X, not the 5950X.

Also, those settings are bone stock with JEDEC timings and all. Compare that to a bone stock 5950x and Intel is faster.

Anandtech scored 1655/15769 in their testing. They run stock with JEDEC timings.

EDIT: If these scores hold up, that means ADL-S is around 10% faster in GB5 than the 5950x, despite being hundreds of dollars cheaper. Zen3D, of course, will reverse things again.
 
Last edited:
moinmoin said:
It would be kind of predictable at this point as AMD does like to increase prices by around the amount of performance added, but a leak claims the highest end Milan chip will cost around 14% more with V-Cache included.

www.tomshardware.com

AMD 64-Core EPYC 'Milan-X' Listed For More Than $10,000

3D V-Cache for more cash
www.tomshardware.com www.tomshardware.com
Click to expand...

Well, that is preety cheep if you take into account the colossal amount of Cache. :grinning:

- if AMD integrates 64mb 3D-V-Cache per CCD

- 64 core Epyc has 8 CCD-s=512mb 3D-V-cache
 
Asterox said:
Well, that is preety cheep if you take into account the colossal amount of Cache. :grinning:

- if AMD integrates 64mb 3D-V-Cache per CCD

- 64 core Epyc has 8 CCD-s=512mb 3D-V-cache
Click to expand...
Imagine how fast 'Puppy Linux' will run, completely in L3 cache(!) (*)

(*) This is a compleletely contrived and hyperbolic example, chosen to showcase exactly how large a 512MB L3 cache effectively is. No-one is actually going to go out and buy a $10,000 USD 64-core server CPU to run a stripped-down Linux desktop OS that already just runs completely in RAM. (Ok, don't give Linus any ideas...)
 
Last edited:
eek2121 said:
I don’t think you realize the full implications of ADL-S. The 12900k is a $500-$600 chip, not an $800 chip. It’s competition is the 5900X, not the 5950X.

Also, those settings are bone stock with JEDEC timings and all. Compare that to a bone stock 5950x and Intel is faster.

Anandtech scored 1655/15769 in their testing. They run stock with JEDEC timings.

EDIT: If these scores hold up, that means ADL-S is around 10% faster in GB5 than the 5950x, despite being hundreds of dollars cheaper. Zen3D, of course, will reverse things again.
Click to expand...

O I do I'm just not impressed with it considering AMD's offerings are a year old. All of those scores on the geekbench site are low for Zen 3. Geekbench is an app that favors intel so I will still need to see more real world benchmarks.

You are impressed with it because you are an intel guy that much is obvious. And there is nothing wrong with that.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top