AMD's Upcoming Zen 4 Based Genoa CPUs Confirmed to Have 1 MB L2 Cache per Core

[mama]

ive been tempted to get a 3700X since im planning to get a gpu, b550+3700X seemed pretty sweet but decided against it

Wanted to get a current gen platform for a while, ddr4+ryzen prices are pretty good (5700X incoming too) but ipc improvement from zen 4 is too substantial to ignore, especially considering it would come along with the feeling of a shiny new platform while also being more future proof with AM5. Oh, and pcie 5.0, which i mean it should be beneficial for ssd related stuff.

Im with a legendary 4790k, for 2160p it should be able to get the job done for a little while.

Something im still very undecided is what to do about gpu, im on desperate need of one due to unfortunate events.

So far my candidates are 6800XT and 3070ti

The thing is, i have many concerns:

-shit is overpriced, especially mid-low range (6700 xt, 3060ti, 6600 xt and such)

-amd is supposed to release a refresh soon...

-prices have started to drop...

-intel gpus coming relatively soon, which should further lower prices worst case scenario, best it outperformens 6800xt and 3080 with a lower price point

-well, if all that wasnt enough, the next gen is dropping this year and should be a massive improvement


So wtf should i do? I basically have no gpu atm, my morals completely refuse to even get a 6700 xt because i consider its way overpriced already at msrp. Yes im poor.

6800xt would be my choice if there was avaliability and/or priced adequately. 3070ti would be a perfect candidate atm if it was on msrp, currently 800eur where i live.

Im being this scrupulous because they are almost 2 year old gpu's so overpaying too much would make me feel even dumber when rdna3 drops.

I guess at the very least i have to wait for the AMD refresh (april right?), and probably get a 6850 xt or a 6070ti if the msrp is dropped (assuming there is avaliability which im relatively confident there will be). I have no hope for Intel to release anything before june and there is a chance the wait would been a waste of time forcing me to wait again but for rdna3 or it could be the other way around, im completely lost.

I am in need of hdmi 2.1 and impatient to play 4k

Future proofing is a myth. It's marketing talk for 'spend more than you need to'. Look at what you have. Consider what you want to do. Buy to suit your needs. As has been pointed out, there are bargains to be had. Personally, I think it's hard to go past a stable AM4 platform but I like what I see with Alder Lake, particularly the 12700K. As for the GPU, again buy what meets your needs and look for a bargain to suit. My take on GPU pricing is that the current price drop on current generation products is most likely linked to the introduction of upcoming releases which won't be cheap. Just get what you need and try to get a bargain.

 

[watzupken]

The war with clockspeed, turned into a war with number of cores, and has now turned into a war with cache size. AMD seems to be very obsessed with increasing the cache size of their CPU and GPU line. I feel this will be useful for now, but may at some point run into the law of diminishing returns. Like this massive increase in L2 cache may negate the benefit of the 3D cache, unless one is running some very cache/ memory latency sensitive application or load. I am nowhere near an expert in this area, but this is just my opinion.

Future proofing is a myth. It's marketing talk for 'spend more than you need to'. Look at what you have. Consider what you want to do. Buy to suit your needs. As has been pointed out, there are bargains to be had. Personally, I think it's hard to go past a stable AM4 platform but I like what I see with Alder Lake, particularly the 12700K. As for the GPU, again buy what meets your needs and look for a bargain to suit. My take on GPU pricing is that the current price drop on current generation products is most likely linked to the introduction of upcoming releases which won't be cheap. Just get what you need and try to get a bargain.

To me, “future proof” is not a myth. It really depends on the individual’s expectation. Some people can live with their computers for years and years, even when it comes to gaming. For example, we see GTX 1060 as still one of the most used GPU today, and is a testament of it being future proof for some. Others expects high or cutting edge performance all the time, so there will be no future proof system. The moment they see something better, they will jump onto it.

 

[Minus Infinity]

Moore's Law is dead broke this news 6 months ago.

Zen 4 will be a huge uplift over Zen 3. This is the biggest architectural change to Zen so far. 25-30% IPC uplifts, 2x L2 caches, higher clocks, possibly 3D cache on many models.

 

[stimpy88]

Interesting that they doubled L2 cache, yet kept L3 the same as Zen3. AMD know that 64MB+ (96MB in the case of the 5800X3D) is what gets the gaming performance they advertise with the 3800X3D. So I will expect to see an upgraded Zen4 with 3D cache, or higher amounts of L3 in Zen 5.

But it's about time the L2 got some love, 512Kb is just not enough. Well done AMD.

 

[RyzenMaster.sys]

I find it funny that people are excited over a chip having 1/3 as much L2 cache per core as we had with late core 2 duos nearly 15 years ago.

Sure those didn't have L3 but still shows that it's possible.

 

[Punkenjoy]

I find it funny that people are excited over a chip having 1/3 as much L2 cache per core as we had with late core 2 duos nearly 15 years ago.

Sure those didn't have L3 but still shows that it's possible.

It was possible indeed. Both AMD and Intel could make 16 MB L2 right now. Doesn't mean that it would be the best move. The main reason is if you can use your L2 data quickly enough, that data is best put into L3. L3 Also serve as a way to communicate between core. An important factor to consider those day is core to core communication wasn't a thing to worry about yet. The multithreaded application were rare and multicore CPU were starting. In their next generation, AMD used the L2 real estate to increase the IPC by widening the core and only put 256 KB instead. They then added a shared L3 that is now the common solution these day.

If you have a too large L2, you can take too much time to lookup if what you request is in it, loosing performance. a CPU core can only compute so many data during x period of time. It's not efficient for data to stay into fast cache for a long period of time if it's not being used. It's then best to put it into L3 or memory. L2 cache take a lot of space so if you put too much, you are wasting opportunities to make your core more wide, increasing the performance.

L3 on the other hand sit outside the core and can be increased without impacting too much the core itself, although in the end, you are still using the same silicon space. This also probably why they don't get over 32 MB yet without the 3D V-cache. They probably simulated that wider core are better than 64 MB of L3 cache for the main die.

The war with clockspeed, turned into a war with number of cores, and has now turned into a war with cache size. AMD seems to be very obsessed with increasing the cache size of their CPU and GPU line. I feel this will be useful for now, but may at some point run into the law of diminishing returns. Like this massive increase in L2 cache may negate the benefit of the 3D cache, unless one is running some very cache/ memory latency sensitive application or load. I am nowhere near an expert in this area, but this is just my opinion.

There will always be a war on the low hanging fruits. the MHz wars was when it was "easy" to ramp up that number, then when they got stuck, they witched to core count. Now, except for very specific workload, there is diminishing return to go beyond a certain threshold at the detriment of faster core. So those days, it's cache. But i don't think that this 1 MB L2 cache is such big thing in the end. it's just to keep up with wider cores. The Added L3 cache on top of the CPU is probably more related. After all, Alder Lake already have 1.25 MB of L2.

But this "massive" L2 cache is not that big and it won't reduce at all the need of a larger L3 cache. They don't have the same purpose. The L3 cache is also used to communicate between cores. It's place in the design allow much larger size than L2 that is situated inside the core. On the latest Epyc 77x3, L1 is about 4 cycles latency, L2 is about 12, L3 is about 50 and a trip to memory is about 300+. (And L3 have way more bandwidth than main memory.)

It's all about what you run, how large is the working set, how frequently you access all of it, etc. Just putting all the time a super large cache wouldn't be a great idea right now like putting a big bunch of core wasn't that great of a idea a while ago. There are not that many application outside game, some server workload and scientific application that would benefits with their data set of having a super large L3. The area is better spend into wider core like in the past, i was smarter up to a point to make 4 wide core instead of 8 small.

That will probably change in the future and we will see soon enough CPU that have GBs of L3 cache.