SegaSnatcher
Member
Jesus Christ!!! 3.8Ghz clock speed for the CPU, I was not at all expecting that.
Didn't he say it would of been last tuesday. Maybe E3 cancellation stuff made the news go to this week. One week off. Better correct percentage then all those that said PS5 Feb reveal!It only took him 10 days to be right.
Lockhart still missing in action btw.
I can't find anything before the 18th of December 2020. Can you link the exact quote?Tommy Fisher wins the prize for nailing the Xbox Series X specs:
He said 52CU at 1.82GHz on 12-14-19 and the final specs were 52CU at 1.825MHz.
No other person on NeoGAF nailed the specs so exactly.
Confused, is it a fixed allocation from a single pool of RAM? Part of the appeal of unified memory was that developers could split it any way, this almost reads like there's 6GB of DDR/system memory, but it also says 16GB of GDDR6, in which case why make a fixed pool?
Xbox have seen the memes, it seems
![]()
52 active CUs at 2.0Ghz ~ 13.3TFlops
PS5
52 active CUs at 1.82Ghz ~ 12.1TFlops
XseX
gl![]()
Jesus Christ!!! 3.8Ghz clock speed for the CPU, I was not at all expecting that.
Yeah, Tommy Fisher > all other insiders so far. The guy even had clockspeeds right and we were all laughing him off.So tommy fisher was right regarding to xsx gpu clockspeed and cu count, which pretty much confirms PS5 will be 52cus @ 2000mhz which would result in 13.3tflops.
This would mean the PS5 is 9.5% more powerful in the GPU.
We still dont know what the differences will be with CPU and Ram, so if ps5 has a lower clocked CPU, the differences would count each other out and the consoles would be even on power.
To put the power difference into perspective it would be like if the xbox one had the same memory setup as the ps4 and a 1.68tf GPU.
Games will perform the same.
Its great news both systems are beasts.
I assume they were looking for more bandwidth, two pools give you more concurrent bandwidth. At least that's what I'm getting from the DF vid.
It seems to be only a single pool.I assume they were looking for more bandwidth, two pools give you more concurrent bandwidth. At least that's what I'm getting from the DF vid.
I assume they were looking for more bandwidth, two pools give you more concurrent bandwidth. At least that's what I'm getting from the DF vid.
We still dont know what the differences will be with CPU and Ram, so if ps5 has a lower clocked CPU, the differences would count each other out and the consoles would be even on power.
Just need to remind you things that i said before for the last time\
GPU: PS5>SeX ~10%
RayTracing Accelerators: PS5>X1 (PS5's RT works by using acceleration structures and algorithms much like nVidia's RTX cores)
CPU: Sex>PS5 ~ 6%
Memory: PS5>Sex for the size and bandwidth
SSD: PS5>Sex ~ Over twice as fast (this is the biggest and most noticable difference)
Fan noise (Dev-Kit): Sex>PS5 (not as egregious 1x vs. Pro).
BTW it's up to you believe me or not, but things gonna be clear very soon and you will find out who knows and who knows shit.
Bonus?
Remedy and Housemarque have been in Sony's WWS family for a while now!
Techland belongs to MS now!
Of course can be 13tf,even 16tf with 15tb of SSD,but how much Sony can invest and how much the final price?MS getting cocky gotta be awkward if tommy fisher was right haha
![]()
If that would be the case, pragmatic game developer would simply develop targeting XSeX.This would mean the PS5 is 9.5% more powerful in the GPU.
Thats what ive been trying to say but just got scolded and told basically to go by "OUR MOST" so disregard the title they used and just go by "OUR:"
It seems to be only a single pool.
They are splitting it via page allocation I guess (software).
It's a trade off between cost and bandwidth (at the same 16GB) - they could have gone 16 1GB chips for a 512 bit interface - huge bandwidth, but more cost not just the chips but extra die space for the 'channels' .. or they could have gone 8x 2GB chips for a 256 bit bus - less bandwidth, but cheaper.
The compromise just makes 6GB of the 16GB a bit slower than the rest .. ok when most of it is being used for OS . the rest of the 6GB can be used for less used assets.
Nope
Nope
If you read the Excel spreadsheet you can see they are all for 3584 shaders, absolutely nowhere is it mentioned that the full chip would have 3840 or 4096 shaders, this is a assumption people have made based on previous consoles having binned chips. What we can probably deduce now is that 36 CU is likely the full on-chip CU count for Oberon/Ariel/Gonzalo rather than 40 that many have assumed.
Nope, the testing is 56 CUs, the console is 52 CUs, like what Tommy Fisher stated it to be.GitHub never mentioned 60 CU's total. Just 56 active. They most likely had all 56 active at time of testing.
Hey man, good to see you back!Holy fucking shit I need to apologize to Tommy Fisher, he got the specs correct.
Holy fucking shit I need to apologize to Tommy Fisher, he got the specs correct.
He was banned after he sent a very shitty shopped photo of the DS4 toWhere tommy fisher, he got banned or what?
I would not automatically assume that everything is 100% legit with him. Naming the complete specs nearly down to the smallest detail and lying correctly is an achievement and actually shows that he knows something or knows someone who does. But the wrong pictures of the "DS5" remain questionable.Holy fucking shit I need to apologize to Tommy Fisher, he got the specs correct.
The decompression is done by hardware, so it isn't that misleading.Did anyone comment on the 4.8 GB/s transer speed for compressed data?? I feel it's really misleading. The transfer speed is 2.4 GB/s, if the data are compressed you may transfer even more data than 4.8 GB/s, but that won't be the norm, compressions have variable rates that depend on which type of data is being processed, so we cannot expect it to always reach 4.8 GB/s. I feel like Microsoft really fooled many people today. Maybe they estimated that the lowest ratio is around 2, so the transfer speed would be AT LEAST 4.8 GB/s for compressed data. Otherwise, Microsoft just put out a meaningless number.
I can't find anything before the 18th of December 2020. Can you link the exact quote?
already 553 now...either way thanks ! i hope everyone across the globe will make through this somehow!I love Malaysia, wonderful country. Be safe!
"Prepare to be absolutely blown away by not just revealing over 13 TFlops GPU power
for PS5 in PSM [2] February 2020."
XseX - 52 active CUs at 1.82Ghz ~ 12.1TFlops (RDNA 2.0)
PS5 - 52 active CUs at 2.0Ghz ~ 13.3TFlops (RDNA 2.0)
Custom AMD Zen 2-based CPU (8 cores @ 3.4 GHz)
Custom AMD Navi-based GPU (52 CU @ 2.0 GHz - 13.3 TF - RDNA 2.0)
APU on 7nm+/7nm EUV process, could explain the high clocks
8GB HBM2 + 16GB DDR4 RAM solution? (4GB DDR4 reserved for OS, power efficiency of HBM2 could be reason for higher GPU clocks)
NVMe SSD with custom controller (focus on read speeds, free CPU by integrated decompression)
I think the RAM and SSD parts are actually much much more interesting than the final TF count. We actually know very little about both PS5 solutions, not a lot of leaks in that direction. Maybe Sony is planning something special and tries hard to keep it secret? Thanks for the read if you made it this far.
sorry to all the insiders here but you all take a back seat to Tommy Fisher.
100% correct on XSX with no riddles.
the guy had balls![]()
![]()
9 seconds to load the game is impressive!!
Tommy Fisher said both are RDNA2 and both have 52 CUs, right?
So, both of them are RDNA2 and one of them is confirmed now 52 CUs^^
Tommy Fisher also said EXACTLY 12.1 RDNA 2 tflops, did he not?Tommy Fisher said both are RDNA2 and both have 52 CUs, right?
So, both of them are RDNA2 and one of them is confirmed now 52 CUs^^
STILL "our most powerful console...."MS getting cocky gotta be awkward if tommy fisher was right haha
He'll be back in May.Someone un ban his ass.![]()