Read more.Doubling peak memory speeds, quadrupling density. Hardware expected in H2 2021.
Read more.Doubling peak memory speeds, quadrupling density. Hardware expected in H2 2021.
So no new PC this year.. somewhere next year around same time as now when it all has matured a bit.
This time next year? Not going by previous historic new memory standards adoption rates.
New hardware in 2021, some servers switching towards the end of the year. Consumer hardware in 2022 or 2023.
DDR4's adoption rate was way slower.
DDR4 was finalised in 2012. The first Skylake CPUs are were mostly DDR3 and only later Skylake architecture CPUs (the so-called Kaby Lake architecture) started using mostly DDR4 except for laptops in 2016.
Last edited by kompukare; 15-07-2020 at 03:10 PM. Reason: typo: meant consumer 2022 not 2020!
I have only just upgraded to DDR4. It'll be at least 3 or 4 years before I consider DDR5.
Quote from the article: "Offer a max standard data rate of 6.4Gbps - 50 per cent faster than the official 3.2Gbps max speed of DDR4."
Erm......6.4 Gbps is 100% faster than 3.2 Gbps.
Just because 3.2 Gbps is 50% slower than 6.4 does not make the reverse true!
"In a perfect world... spammers would get caught, go to jail, and share a cell with many men who have enlarged their penises, taken Viagra and are looking for a new relationship."
Sumanji (15-07-2020)
I was originally thinking about a big upgrade this year, then changed my mind to next year when DDR5 and the new platform is released, but I'm back to thinking this year again... as you say, DDR5 launch unlikely to happen next year for consumers, and I'm sure the new memory and mobo's will be expensive, not to mention new adopter's teething issues. Zen4 maybe an end of the line platform, but it should provide many years of solid performance.
But neither Intel or AMD make memory chips.
And those who do (44% Samsung, 29% Hynix, 20% Micron) operate on often razor thin margins unlike logic silicon vendors.
Unsure who much retooling will be required for DDR5 but wasn't it that usually they don't retool but rather the new fabs at smaller nodes will be designed to DDR5 while the old ones continue with their current output.
Either way, I'm sure the top vendors will be happy to get higher prices for DDR5 if the demand is there. Unless one of the Koreans tries to bury Micron (who by all accounts are significantly less efficient), but why risk the ire of Trumpster?
When AMD or Intel adopt DDR5 is not that relevant to those calculation except to drive demand which if it outstrips supply will just mean those industries less dependent on price will adopt it first. Like for example servers. Which is happened at the previous switch and probably the switch before that.
Finally.
Traditionally at the point of crossover AMD and Intel have made parts with dual standard ram controllers which can take either memory type. So AMD could make some AM4 parts re-released on AM5 giving a performance boost.
The split channel per dimm will be useful for APUs and to help feed the multiple threads on modern PCs. It also makes it rather more like GDDR memory, so no doubt we will see low end graphics cards adopting it given we saw DDR3 gpus where the hit was pretty bad.
AMD could just swap out the IO die, saving some silicon area
That's an interesting thought, given how DDR5 seems to have slipped a bit. Perhaps just needing a new IO die for AM5 was part of the reasoning for the 3000 series, though seeing how fast the new APUs are I wonder if 4000 series will be dropping chiplet and going back to fully integrated monolithic dies.
Anxious to build September yet DDR5, PCIe5, USB4 & 5nm suggest 2021 may be prudent. Prob throw Ampere/Sienna Cichlid in my antique Haswell till then.
There fixed that for you!
Certainly for affordable DDR5 - and for that matter DDR5 which actually offers a tangible improvement on current top-of-the-range DDR4.
PCIe 5.0 and USB 4.0 might be possible but PCIe 4.0 is currently mostly under-utilised (headline burst rates for NVMe drives are about as useful headline burst rates for SATA3, SATA2, SATA1, PATA133, and so on - that is not very).
5nm should happen in 2021, but not for everything as the prices of designs, masks etc. are even worse than 7nm. From the performance of the coming Zen2 APUs it is clear that a monolithic design has many advantages, and while AMD's increasing volume means it would make sense for them to return to monolithic for mainstream, the increased costs at 5nm make that very unlikely.
There are currently 1 users browsing this thread. (0 members and 1 guests)