VolticArchangel
Member
GAF > Internet > GAF
The funny thing is he posted that almost a month ago and I just got to see it thanks to that sub early in the day lol
The way we are currently heading, if you extrapolate doubling every 1.5 years, (recently looks like we are accelerating beyond that again), you'll be able to far surpass PS5 on a mobile device by 2029.PS4 Pro (GPU + Memory Bandwidth; along with efficiency gains with architecture & ML)
PS5 base (CPU + Storage I/O + RAM)
I'd be looking at early PS5/late PS4/cross-gen visuals rendered at 720p @ 40-60fps, with 2-3x frame gen + 2x ML upscaling to 1080p @ 120fps.
Mainly make it a legacy device to play everything from PS1 to PS4 Pro + PSP & Vita natively excl. PS3, then streaming all of the same incl. PS3, having local Portal-like playback for PS5 & PS6. Having some native mid budget titles developed for it and then porting down select PS5-only titles. I'd also see if I can entice over some third party Switch/Switch 2 stuff. Then throw in a basic android sandbox for mobile titles and additional functionality.
One thing's for sure, I don't want PS6 games being developed with any handheld in mind. I want that thing pushed to the limits, not being hamstrung by something with <1/15th the power draw. There's only so much scaling you can do before entire featuresets just stop becoming viable and you end up having to make two functionally different games; which they probably won't wanna do.
Laptop GPU performance isn't doubling every 1.5 years though. It's not clear if we have even doubled mobile 3080 Ti performance yet.The way we are currently heading, if you extrapolate doubling every 1.5 years, (recently looks like we are accelerating beyond that again), you'll be able to far surpass PS5 on a mobile device by 2029.
As the die continues to shrink, at some point it won't much matter what kind of shell you put around it. Sometime next decade, your "tablet" should have very little performance difference than a tower PC, especially when we move beyond silicone.
I'm sorry but efficiency can't make the rtx 6010 runs like the 5090. That equals to Jensen Lies to the power of 10.
Can a PS6 Portable Be as Powerful as a PS5 by 2028?
1. CPU (Processing Power) – Yes, Likely
- PS5: 8-core Zen 2 CPU @ 3.5GHz (~448 GFLOPS CPU power)
- PS6 Portable (~2028 Expected): 8-core Zen 5/6 CPU @ 3.0-3.5GHz (~600+ GFLOPS CPU power)
Verdict: Yes! By 2028, portable chips will be much more efficient. An 8-core Zen 5/6 CPU could match or outperform PS5's Zen 2 CPU while using far less power.
2. GPU (Graphics Power) – Possibly, Depending on TDP
- PS5: 10.3 TFLOPS RDNA 2 GPU (36 CUs @ 2.23GHz)
- Steam Deck (2022): 1.6 TFLOPS RDNA 2 GPU (8 CUs @ 1.6GHz)
- ROG Ally (2023): 3.8 TFLOPS RDNA 3 GPU (12 CUs @ 2.7GHz)
- PS6 Portable (~2028 Expected): 6-10 TFLOPS RDNA 5+ GPU (~20 CUs @ 2.5GHz?)
Verdict: Maybe.
- A 6-10 TFLOPS GPU in a 30-50W handheld is possible with RDNA 5+ or AI-assisted upscaling (FSR 4.0, AI reconstruction).
- But it won't be exactly 10.3 TFLOPS like PS5 unless Sony pushes higher power consumption (50-60W docked).
3. RAM (Memory & Bandwidth) – Likely
- PS5: 16GB GDDR6 @ 448GB/s
- PS6 Portable (~2028 Expected): 24GB LPDDR6X @ 400-500GB/s OR LP-HBM @ 800GB/s
Verdict: Yes!
- LPDDR6X (expected ~2026) could provide 400-500GB/s bandwidth, nearly matching the PS5.
- If Sony uses LP-HBM (~800GB/s), it would surpass PS5 in memory bandwidth!
4. Storage – Faster than PS5
- PS5: 5.5GB/s SSD (9GB/s compressed)
- PS6 Portable (~2028 Expected): PCIe 5.0 SSD (~10GB/s)
Verdict: Yes! A PCIe 5.0 SSD in 2028 will be at least 2× faster than PS5's storage.
5. Power Consumption – Portable vs. Console Tradeoff
- PS5: ~200W+ TDP (console)
- PS6 Portable (~2028 Expected): 35W-50W TDP (portable mode), 60W+ docked mode?
Verdict: It depends on power limits.
- In handheld mode (~35W-50W), the PS6 Portable may run at ~6-8 TFLOPS.
- In docked mode (~60W+), it could hit PS5-level performance (~10 TFLOPS).
- AI upscaling (FSR 4.0, DLSS-like tech) can compensate for lower raw power.
Final Answer: Will PS6 Portable Match PS5?
Component PS5 (2020) PS6 Portable (~2028 Expected) Can it match PS5? CPU 8-core Zen 2 @ 3.5GHz 8-core Zen 5/6 @ 3.0-3.5GHz Likely better
GPU 10.3 TFLOPS RDNA 2 (36 CUs @ 2.23GHz) 6-10 TFLOPS RDNA 5 (~20 CUs @ 2.5GHz?) Maybe (~6-8 TFLOPS portable, 10 TFLOPS docked)
RAM 16GB GDDR6 @ 448GB/s 24GB LPDDR6X (~500GB/s) or LP-HBM (~800GB/s) Likely equal or better
Storage 5.5GB/s SSD PCIe 5.0 SSD (~10GB/s) Likely faster
Power (TDP) 200W+ 35W-50W (portable), 60W+ (docked) Depends on TDP
Final Verdict:
A PS6 Portable in 2028 CAN match or exceed the PS5 in most areas, but it depends on power limits.
Possible Performance Modes:
——————————————————
Portable Mode (~35W-50W): 6-8 TFLOPS (PS4 Pro+ to near-PS5 performance)
Docked Mode (~60W+): 8-10 TFLOPS (matching PS5 with better RAM & SSD)
Full PS5 power (10.3 TFLOPS) in handheld mode? Unlikely, unless power efficiency drastically improves.
By next-gen, we will have cross-gen games and probably 99% of games will be built with PS5 as the least common denominator. With that in mind, PS6 at 50w tdp at $449 that is small enough to be portable and can be attached to a separate sort of PS portal machine (controller+screen+battery) would be a great idea! PS6 base will be cheap and PS6 Pro will be the 200w-300w monster machine.
It may be a possibility that it may not benefit from any AI hardware.They are still on the power hype train. This device will have RDNA4 RT and FSR4 to target PS5 Image quality. Remember that it's likely been designed by Cerny and that PS5 had shit RT and no AI uspcaling.
Can't be any better releasing in 2028 that 2027, if design is already complete.I'm hoping PS6 can be 2028 so it's the best it can be.
PS6 is design complete and in pre-si validation already, with A0 tapeout scheduled for late this year.
It may be a possibility that it may not benefit from any AI hardware.
Seems like it requires a certain amount of GPU power to do upscaling.
Also, there's no RDNA4 APUs in the near future.
They had better let developers decide if a game releases on both consoles or not to avoid the Series S/X disaster.The performance gap would be too large if they are attempting some sort of mandatory dual tier platform like the Series S|X. You'd have to pare down your main console to make it work...and I don't see Sony doing that for PS6.
This is the first time you've acknowledged a handheld PlayStation SoC, or did I miss anything?No, it's a 15W SoC on 3nm.
Is this AI generated garble really worth sharing?
The performance gap would be too large if they are attempting some sort of mandatory dual tier platform like the Series S|X. You'd have to pare down your main console to make it work...and I don't see Sony doing that for PS6.
They had better let developers decide if a game releases on both consoles or not to avoid the Series S/X disaster.
Not at 15w.The latest high end iGPU from AMD the 8060s in the Ryzen AI MAX+ 395 is roughly 21% slower than the RX 6700 which is very close to the GPU of the PS5. This is still using RDNA 3.5, switch to RDNA 4 or UNDA might make up the performance difference at the same CU counts.
https://technical.city/en/video/Radeon-RX-6700-vs-Radeon-8060S
If Sony ever releases a mobile companion device, i'm pretty sure they would never allow supporting it optional.Not a chance.
PS6 Portable will be much like Steam Deck, optional for games that support it, but not a requirement. May support base PS4 games universally without patches, but not PS5.
Not at 15w.
Seeing Strix Halo at 15w made me wonder if Sony can repurpose PS6 chips with less usable CUs for the handheld.Is that at 1080p? Running that game at just under 60fps while completely chocked on power is very impressive. Especially if that is native.
According to DF, Returnal's native resolution is only around 1080p. Housemarque says they then used temporal upsampling to get to 1440p and then checkerboard rendering to get to 4K
That of course depends on all of the other graphical settings. I didn't see anything linked in the Tweet.
Not surprised.
the only way to make a Portable PS5 by 2028 is if it's bigger than a Steam Deck, costs 2000€ and has a battery life of 30min at full load...
is it technically doable? sure, you could do it today in fact... would it be an appealing product? absolutely not due to physical restrictions
Great, now AI is taking our moronic drunk/high theory threads. You don't even have to write your own nonsense scenarios anymore.
We used to be a proper forum, I tell ya.
Sure seems like its comingThey NEED to cancel this thing. It will ruin the gen.
I'm not sure if AMD can test the CPU or GPU before it's merged on the die with the rest of the SOC. I know they can't for the X3D chips in regards to how the memory is connected.Seeing Strix Halo at 15w made me wonder if Sony can repurpose PS6 chips with less usable CUs for the handheld.
Could that reduce the overall cost of both the handheld and console?
Sodium Ions are already being worked on, similar volume capacity as LiON on the low end. But more than twice the capacity in W/Kg.Someone ask Chatgpt to create a new battery while they're at it.
To understand what I meant, take a look at the PS5. It has 40 total compute units, but 4 is disabled to increase yields.I'm not sure if AMD can test the CPU or GPU before it's merged on the die with the rest of the SOC. I know they can't for the X3D chips in regards to how the memory is connected.
AMD has been very good about recycling parts that don't meet their spec goals the past few generations.
People still don't get this… It works for Nintendo because is an hybrid. Sony is not going to fallow that route.The performance gap would be too large if they are attempting some sort of mandatory dual tier platform like the Series S|X. You'd have to pare down your main console to make it work...and I don't see Sony doing that for PS6.
A single handheld that uses the exact same games as a TV counterpart.Sony won't be creating their own Wii U. It will be a single handheld.
Silicon-carbon batteries would be the solution to this.40 watts in handheld mode… they need a new tech for batteries to play more than 1 hour or a really bulky device.