Next Generation Hardware Speculation with a Technical Spin [2018]

Status
Not open for further replies.
I dont see Sony leaving behind the huge PS4 installed base... Its probably a ps4 super pro coming... High bandwidth16 giga GDDR6, maybe 16 jaguar cores... ok 8 cores are going mostly to be dead silicon
 
what about implementing on new and old hw (via firmware update) a special criptocurrency mining "sleep mode" with wich to pay then new videogames and play station plus ?
 
imo the leap in TFlops will depend on how AMD will improve their perf/watt metric with their GPUs...
Current Vega's perf/watt isn't spectacular to say the least.

EDIT:

With what AMD has currently, I wouldn't expect much more than 10TF.
 
In Nov 2013 the PS4 had ~ a $170-180 GPU. Equivalent performance to a slightly overclocked 7850.** By the time of PS4's launch in Nov 2013, you could get a HD7850 for $170-180 fairly regularly. (PS4/X1 gpus are from the same architecture family (Southern Island to be specific as the HD7850))

Do we really think AMD will have 12-15 tflop videocard for $170-200 MSRP in 2 years? Look at GPU prices prior to the mining craze, we are asking for a HUGE performance increase in 2-3 years.
Lets go to $200 what can that get you in pre-cryptocurrency inflated prices.... a GTX 3GB 1060 (4.2 TFLOPS), and a RX470 4GB (4.9 TFLOPS) ....

At a 399 MSRP of PS5/X2 the GPU inside is going to be equivalent to a $200. Unless they go for a $499 MSRP like the XOX. Do we really expect a $200 videocard in 2020 to be 12-15tflops?

** technically- (lower clocked than the stock 7850 and 7870 but with more shader/texture units than a 7850 but less than a 7870)
 
Last edited:
yes pixel you are right... Thats why silicon for cpu will be kept at minimum (shrink of jaguars at 7 nm) and maybe a 2xpro 8.4 TF GPU can also be used (on "standby") to mine out criptocurencies so the console owner can pay new videogames and more with the "profit"... A new business model.
 
I more and more think that with their Premium and base model, they can do a super-premium model with their PS5, selling it at 500€.

Otherwise they'll get stuck at ~10tflops no doubt with a 400€ tagprice. And honestly prices haven't moved much from 2013. More than 4 years after release they still sell their PS4 300€ and now Pro at 400€.
 
They're either gonna have to take a heavier loss at launch for $399 or increase the price to $499 this time, I'm sure of it. It will be increasingly hard to differentiate graphically from the mid gen refreshes if not enough power is there to boost the leap. Can we say mid gen upgrade is a double edged sword now? Man the odds are really against Sony and MS this time isn't it? They need 4 times the resolution 4k vs 1080p as opposed to 1080p vs 720p on top of a generation graphics leap all selling at a reasonable price and fit inside a console box? Quite a tall order I'd say. Sure they could wait longer maybe launch at late 2021 but I don't see Sony waiting that long.
 
I think checker board rendering will be standard next gen, no need to foolishly waste resources on native 4K. The techniques for this will only improve in the future.

I also don’t really see what a Pro model will really offer. 1080p to 4K was a big step that won’t be repeated. 4K checkerboard to native is negligible.
 
For the planning of the next consoles its relevant what AMD has in the pipeline for 2020-2021 where much is already finished and thats why Vega is irrelevant. No shared memory pool slows down engines, makes it more inflexible and backward compatibility difficult as well.

8 cores (minimum)
64GB
12 Teraflops

Memory is always a problem. Most of the production is looking for storage by all means. In addition, UHD needs huge render targets and actually a mip more on all textures. The PlayStation 4 Pro has problems with it. Also with enough computing power one wants for example a clever voxel GI and that needs a lot of memory. Moore's Law has not slowed down with memory as far as I know. Memory prices are high because of the demand while the technology advancement is not yet an issue

Neural networks will also be used in the future to create better and faster anti-aliasing, depth of field, screen space reflections, SSAO etc.
Studios have a strong interest in reducing content costs of AAA graphics, shortening iteration times etc. There will be some exciting SIGGRAPH presentations in the future.

I think checker board rendering will be standard next gen, no need to foolishly waste resources on native 4K.

No.
 
Last edited:
If they release at $500 how will they get the price down to $300 within three years?
Probably unlikely if the major cost ends up being RAM, which would have had some special deal for volume supply in the first place.

For the APU, they'd only be able to rely on better yields. It'd be tough to launch with a break-even device while pushing everything. I mean, is the included storage tech going to remain just as feeble as it is today?

They can mitigate some APU costs with binning multiple tiers ala Pro/XT GPUs (disabled units, lower clocks) just to avoid throwing away good-but-not-up-to-spec chips.
 
Capacity is being built up a lot because one can make a lot of money. Then the price falls again for a few years and then it goes up again. Its always the same.

There has always been a boom/bust cycle in the DRAM market, and every bust saw the price plummet to a much lower level than the previous bust, but not anymore.

Silicon scaling is dying and with it "free" large capacity upgrades. Expansion of capacity now involves building new fabs, which is more costly and takes longer. The past 18 months are almost identical price-wise with the last boom in 2012 to late 2013. Current $/GB DRAM is almost the same as in 2008 !!!

Current spot price for DRAM is around $10/GB. Even if it falls back to $3/GB, you'd still be sinking $200 in DRAM alone on your 64GB concole. Not going to happen!

Cheers
 
Probably unlikely if the major cost ends up being RAM, which would have had some special deal for volume supply in the first place.

For the APU, they'd only be able to rely on better yields. It'd be tough to launch with a break-even device while pushing everything. I mean, is the included storage tech going to remain just as feeble as it is today?

They can mitigate some APU costs with binning multiple tiers ala Pro/XT GPUs (disabled units, lower clocks) just to avoid throwing away good-but-not-up-to-spec chips.

Seems pretty obvious that there will once again by tiers of performance. Base and a Plus. They need the base model to go from $400->300, and then a Plus model with high price, that they don't need to worry about reducing, to maximize revenues.
 
Seems pretty obvious that there will once again by tiers of performance. Base and a Plus. They need the base model to go from $400->300, and then a Plus model with high price, that they don't need to worry about reducing, to maximize revenues.
It would be interesting to see how they'd approach it if that's how it goes. The vast majority of people will be upgrading from the 2013 base spec, so even a 9TF/Ryzenator as the new base would still be a leap. Meanwhile, they can push the higher end SKU @ 12-15TF for early adopters.
 
The vast majority of people will be upgrading from the 2013 base spec, so even a 9TF/Ryzenator as the new base would still be a leap. Meanwhile, they can push the higher end SKU @ 12-15TF for early adopters.

I think there are a lot of arguments for keeping the core SOC + DRAM constant; economies of scale, simpler validation etc. Then differentiate on accessories.

Maybe a flash-only base SKU (Arcade, yay!) and a premium model with additional HDD+optical storage + possibly extra/elite controller. Launch at $300 for the base unit (loss leading) and $400 for the premium unit (at a profit). Drive the price down to $200 / $300 for base/premium models in three years time, then launch a PRO/X upgrade at $400.

Cheers
 
8 cores (minimum)
64GB
12 Teraflops

64GB seems like a lot to ask. GDDR6 is expected to go upto 32Gb per chip, so we may see 32GB (8 chips / 256bit bus) or 48GB (12 chips / 384bit bus). With HBM, HBM3 will have densities that could support 64GB with 4 stacks, I don't think HBM3 will be ready in 2 years for consumer deployment nor would they put 4 stacks on the interposer for a console.



Yes. :)

XBX1 can do 4K native for some games with 6Tflops. 12 Tflops would probably be sufficient for native 4k with minimal improvements overall in rendering. I don't think that will sell. There needs to be a compelling visual reason to upgrade.
 
Maybe a flash-only base SKU (Arcade, yay!) and a premium model with additional HDD+optical storage + possibly extra/elite controller. Launch at $300 for the base unit (loss leading) and $400 for the premium unit (at a profit).
That's been proven super unpopular before, no? XBox arcade and 20GB PS3s weren't at all popular. Early adopters want the full-fat experience, with gimped consoles only making sense down the line selling to the bottom end. A normal console and expensive elite makes more sense to me, at maybe $400 and $600. You're bound to get your biggest fans paying the upper price.
 
Status
Not open for further replies.
Back
Top