Playstation 5 [PS5] [Release November 12 2020]

What I take from this video is that even without heatsink the NVMe tops out at ~53ºC when doing sequential reading, and NVMes seemingly only start throttling at close to 70ºC.
It seems to me that people who buy the add-in NVMe for the PS5 to actually play games on it don't really need to get any heatsink on it.
He is measuring the NAND flash chips temperatures and not the SSD controller temperature. Performance throttling is related to controller temperature not NAND chips temperatures
 
What I take from this video is that even without heatsink the NVMe tops out at ~53ºC when doing sequential reading, and NVMes seemingly only start throttling at close to 70ºC.
It seems to me that people who buy the add-in NVMe for the PS5 to actually play games on it don't really need to get any heatsink on it.
His argument that waffling over the cost of a <$20 heatsink for a >$200 SSD seems pretty sensible given the temperature differences. Why not run 4-8° (idle, write vs read) to 15-20°C (sustained read vs write) cooler? Not to mention the temp deltas were only ~5-6° with a heatsink vs ~14-20 without. He mentioned measuring ambient alongside NAND temp, but I don’t remember him saying what it was or if he tested enough to see how they correlate in a stifling room.*

* I see 23°C (73°F) at 6:23. That’s not exactly a hot room.

Edit: I couldn’t find an associated article at that YT vid’s website, nascompares.com. I did find the PS5 Internal SSD Expansion Megathread on Reddit (better compatibility and heatsink info table here), though no obvious temp data there.
 
Last edited:
Where im 23 celcius is far from a hot room. The high 20's is considered a hot room, then the temp usually is higher under the tv along other electronics.
 
His argument that waffling over the cost of a <$20 heatsink for a >$200 SSD seems pretty sensible given the temperature differences. Why not run 4-8° (idle, write vs read) to 15-20°C (sustained read vs write) cooler? Not to mention the temp deltas were only ~5-6° with a heatsink vs ~14-20 without. He mentioned measuring ambient alongside NAND temp, but I don’t remember him saying what it was or if he tested enough to see how they correlate in a stifling room.*

* I see 23°C (73°F) at 6:23. That’s not exactly a hot room.

Edit: I couldn’t find an associated article at that YT vid’s website, nascompares.com. I did find the PS5 Internal SSD Expansion Megathread on Reddit (better compatibility and heatsink info table here), though no obvious temp data there.

Yes, I agree that getting a cheap 3rd party heatsink won't do any harm and that's probably what I'll do in the long run. But from those results I see no reason to pay the ~$50 heatsink premiums over the otherwise $180 heatsink-less 1TB models.
Especially considering the drives will still be working well within spec in 99% of the console's typical usage even if they don't have any heatsink, and they won't throttle and cause performance issues.
 
Any word from Sony on increasing availability?

Or are we going through another fall and winter with chasing down a few units here and a few units there?

Honestly, they should just accumulate a big number to dump it in the stores in November and December.

Make it seem like there's no more scarcity, to at least try to discourage the scalpers.
 
So DLSS is also useless because it is using 16k images to train its engine. They should just train against 4k images.
Yeah... right.

Doing 8k in the touryst, or 16k in REZ aint all that exciting neither surprising. They probably can do 32k for the first doom but whats the point.

Meanwhile were still 1080p/1440p for native next gen games, topping at 60 or even 30fps.
 
Doing 8k in the touryst, or 16k in REZ aint all that exciting neither surprising. They probably can do 32k for the first doom but whats the point.

Meanwhile were still 1080p/1440p for native next gen games, topping at 60 or even 30fps.
I ask this, is DLSS training against 16k is useless? Is a game supersampling to 8k is useless?
Your post didn't say it was unsurprising. You basically saying supersampling on a non demanding game is useless. Basically your post simply want to downplay this game that just happen to do 8k supersampling on PS5. If this happen on PC or XsX you probably won't say anything (btw, they also do supersampling on XsX iirc, 6k on XsX and 4k on XsS).
 
Doing 8k in the touryst, or 16k in REZ aint all that exciting neither surprising. They probably can do 32k for the first doom but whats the point.

Meanwhile were still 1080p/1440p for native next gen games, topping at 60 or even 30fps.
I would love to play Doom in software mode with alias free point sampled textures. (Most likely a nice antialiased textel edges and mipmapping would be quite bit cheaper..)
 
Last edited:
I ask this, is DLSS training against 16k is useless? Is a game supersampling to 8k is useless?
Your post didn't say it was unsurprising. You basically saying supersampling on a non demanding game is useless. Basically your post simply want to downplay this game that just happen to do 8k supersampling on PS5. If this happen on PC or XsX you probably won't say anything (btw, they also do supersampling on XsX iirc, 6k on XsX and 4k on XsS).

REZ at 16k was/is done on a pc. It aint all that inspiring. Neither is Touryst at 8k (native or not). Its just for checkmark filling.

Whats next, mario at native 8k?

I would love to play Doom in software mode with alias free point sampled textures. (Most likely a nice antialiased textel edges and mipmapping would be quite bit cheaper..)

Doom 3 at 16k

 
REZ at 16k was/is done on a pc. It aint all that inspiring. Neither is Touryst at 8k (native or not). Its just for checkmark filling.

Whats next, mario at native 8k?



Doom 3 at 16k

Nice.

Pretty sure original D3 already did work with SGSSAA and such tricks, so through tweaks one could run it ~32xSSAA with better sample placements on NV hardware. (basically 4x resolution scale + 8xSSAA with MSAA pattern.)
For original Doom I love low resolutions though, maximum quality at 320x200 or 640x480 max. :D

I wonder what tricks are now a days plausible for getting most quality for performance when shading rate is at one sample per pixel..
Something like 4xMSAA buffer where shading subsample is changed per frame or even more often, let TAA handle the jitter or noise. (Or larger than frame ID buffer etc.)
 
I'm confused, PS5 can't output 8k? I though it can output 8k

Or the tweet just says that for those plebs with 4k or 1080p TVs. The game will still render at 8k as long as 60 fps mode is activated?
 
REZ at 16k was/is done on a pc. It aint all that inspiring. Neither is Touryst at 8k (native or not). Its just for checkmark filling.

Whats next, mario at native 8k?



Doom 3 at 16k

Is it useless tho (as in doing nothing)? Basically you quick to say that it is useless since you really quick to post something with negative tone about PS5.
Personally I don't think it is useless but at the same time I don't think it is anything special for that game to be able to do it.
If this kind of thing is just a checkmark thing (enabling something just because it can/has thr headroom to do it) then I can list lots of thing that fall under this checkmark feature thing.
 
I'm confused, PS5 can't output 8k? I though it can output 8k

Or the tweet just says that for those plebs with 4k or 1080p TVs. The game will still render at 8k as long as 60 fps mode is activated?
Pretty sure output to 8k is not yet enabled.
In theory the HDMI spec should allow it.

Game will most likely render a some form of 8k image and downsample it to 4k. (4x SSAA)
 
Pretty sure output to 8k is not yet enabled.
In theory the HDMI spec should allow it.

Game will most likely render a some form of 8k image and downsample it to 4k. (4x SSAA)
I wonder why they dont enable it.
They market it on the box.
Btw can the PS5 funtion as a media player that can play your video and audio files through USB?
 
Back
Top