don't! Buy system today, flip it when ddr5 comes out, get another then. That solves the problem wait or buy now that you always have in the tech space.
@Jan Hendrik Franke In the video mid/late 2022 was speculated. Surely there would come out something eles to fill the gab. Maybe something like Zen 3+.
@@BrandonYen even with it coming out this year, I still would want to wait for the next cpu after that. Mostly because 1st gen/new platform & they can iron out all the bugs. And based off the last transition from ddr3-ddr4, the high end ddr3 was better & cheaper than the first gen ddr4. So I will probably wait a year after ddr5 platform comes out.
@@BrandonYen I really hope so for them, otherwise I believe their processors will not catch up with apple. Unless they integrate powerful accelerators that is.
There will likely be a Zen 3+ in 2021. This Zen 4 release in Q3 2022 would fit that and it wouldn’t be “late”. There are already indications of a Zen 3+.
Everyone should expect that chip shortages will last through the release of zen 4 and rtx4000 even if there is higher supply because of the un-filled demand for the equivalent products from a year or two before
I don't trust that one. Wouldn't be surprised if it's just either two or three CCDs, with the remaining additional dies being either HBM stacks or some iGPUs.
I really REALLY hope there is a big delay in everything noteworthy. I wouldnt mind if Zen 4 and RDNA 3 isnt a thing before LATE 2022 like november-december or even early 2023. CPUs and GPUs are released way to frequently. You get new hardware thats super hyped up and then in literally just a couple months its already outdated and considered "last gen". Thats stupid, this stuff is expensive. Just the fact that we are already talking about next gen stuff is ridiculous, the current stuff only just released! Us (consumers), we let these companies get away with far too much.
I don't see that much improvement coming from 4 threads per core for now. Multithreading as it has always been is sort of like a bandage to help keep the cpu core fed (to allow flexibility) and not leave execution units idle that could be working. There are better ways to keep the cpu fed than just adding more threads. A wider decoder (maybe) and larger reorder buffer(L4 cache), I assume, would be better than adding more threads per core. With the beefy cache, I don't see how more threads would add that much. Overall looks promising. :)
beefy l4 cache sounds promising. remembering broadwell being praised for its l4 when it came to pushing very low latency frames in gaming and overall responsiveness
Brandon , you are wrong avout the IPC...it's not from Zen...because there already is about 35% IPC increasement between Zen and Zen 3. So it looks like they want just to burry Intel...and it seems like the IPC improvement is compared to Zen3! Otherwise it makes no sense. Zen 3 had about 19% over Zen 2, that had about 13 to 15%...and Zen + was only a few percent as it was just a refresh.
If AMD waits until Q3 2022, ;they are leaving the door open for Intel to get back into the game. Intel isn't sitting still during all of this. One of these days they are going to surprise the market with something competitive. Zen3 is fantastic, but not insurmountable. 2 years gives Intel too much time to mount a competitive counterstrike strategy.
AMD believes Intel will have a competing 7nm product in 2022, so that's why they are also able to delay their release a little longer and improve the chips so they can make sure that their release will beat out Intel's
There is supposedly going to be zen3 refresh (with some very promising gains in frequency, allegedly). This should act as a good stop-gap for AMD and overshadow Intel's 11-series pretty badly (although I expect it won't be for a while yet). Go research codename "warhol" and you should get lots more info on this, even though it's all leaks and rumors at this point.
This leak seems questionable to me. There are just too many inconsistencies... 6600MHz is just an unlikely target clock speed for example. All the DDR5 standard clock speeds are multiples of 400MHz. 4 CCDs just sounds very unlikely. 2x12 Core CCDs is what the rumour mills are currently predicting.
As I said this is complete speculation (from our perspective) from an AMD employee. As for the 400mhz clock speeds, those only go up to 6400mhz, so it could be completely possible that we would see smaller increments.
@@BrandonYen That’s absolutely fair... Afaik there is nothing technical preventing 6600MHz, it just perhaps isn’t the most conventional clock speed out there.
2:40 2:1 DRAM to FCLK with a 3300 FCLK is not 6600mhz, that'd be 13200mhz, 1:1 3300Mhz fabric would be 6600Mhz mem, remember, DDR = double data rate, which means it's IO clockspeed is half of its effective transfer speed.
Sorry, I should have mentioned I was talking about MT/s when referring to DRAM instead of UCLK. Should have probably used what you mentioned. Thank you!
@@RobBCactive Yes but 2:1 ratio, the 2 here being DRAM and 1 being FCLK means 2x dram clock per 1x FCLK, 1:1 FCLK would be 6600, but 2:1 would be 13200.
Honestly, intel was always the price to performance king, (for gaming) it’s just how they launch and change things, “oh boy can’t wait for coffee lake ++++++++ Xtreme edition with their l3900 socket
Never underestimate Intel. They know how to build cpu's. And Intel kinda woken up finally. If they get the same amount of performance out of 10nm Intel (equivalent to 7nm TSMC) as they did with 14nm, then AMD is completely lost even with 5nm TSMC. Intel is a gigantic company and they hate losing money, so they do everything to wipe the floor with AMD in 2022/23, hope AMD can fight back. That will be good for the consumer and prices.
I feel like their 7nm will be a good competitor but not 10nm. Willow cove/Golden cove are going to be on 10nm+, so just an improvement of Sunny Cove (cores on Ice Lake, which did not clock very well/perform very well). Unless Intel can make a big jump from Sunny Cove, 10nm+ is unlikely to do well against AMD.
DDR5 will be expensive, am5 first cpu will have alot of bugs just like the initial ryzen, it wont be worth whoever buys this will pay a premium to beta test
I don't think this rumour is plausible. Intel is planning to have Alder Lake out in the 2nd-half of this year. If AMD doesn't get Zen 4 out until 2nd-half of 2022, then Intel will soak-up all the demand for the brand-new generation of PC systems which support DDR5, PCIE5, and USB4 / Thunderbolt 3. That would be a huge loss. There's no way that AMD could spin that kind of a head-start for Intel to "beat" intel. Intel would be the first-to-market for the biggest new generation watershed moment and have no competition in that market for almost a year. That would be PR disaster for AMD no matter how good Zen 4 would be a year later. In that case, AMD would not be drowning-out Intel at all - Intel would be drowning-out AMD for a full year because AMD wouldn't have DDR5, PCIE5, and widespread USB4 / Thunderbolt 3 adoption. Also, in 2nd-half of 2022, AMD won't be competing with Alder Lake, but with Intel subsequent CPU. So, this rumour completely misses that the competitor to Zen 4 isn't Intel's post-Alder Lake CPU, but is Alder Lake, even though it will be on 10 nm. The next generational transition isn't about the lower nm, it's about the DDR5, PCIE5, and USB4 / Thunderbolt 3 being together in one package as the new standard.
Yep, AMD has said in 2019 that they were "on track" to release Zen 4 in 2021, but there are a lot of trusted leaks that point to it releasing in 2022. To be fair, the competitor to Zen 4 is both Alder Lake and Meteor Lake (depending on the performance of Alder Lake - if it performs extremely well and is affordable compared to Zen 3, AMD could be in trouble, but this is unlikely). Again, the leak mentions that AMD wants to make sure that Zen 4 is extremely good so that it blows Intel out of the water.
@@BrandonYen But no matter how good Zen 4 is, if it releases a year later than Alder Lake it'll be getting Intel's leftovers after the craze to get DDR5 and the other new standards is mostly over. A year of sold-out sales without any competition on DDR5 platforms would make Intel the winner by default, where sales are concerned.
