SteveGrabowski
Diamond Member
- Oct 20, 2014
- 8,575
- 7,196
- 136
SteveGrabowski
Diamond Member
- Oct 20, 2014
- 8,575
- 7,196
- 136
I think Ratchet & Clank Rift was confirmed to be a 30 fps game sadly. I hate this RT crap, consoles finally will have the hardware to push 60 fps and instead it's going to be a lot of overkill resolution overkill effects 30 fps gaming.
Wow, uhh, no.
First off you are absurdly off on your CPU projections, in ways that don't even make sense.
Best case scenario for upcoming consoles is the Xbox when talking power and that at best hits 3.8GHZ, we have several thousand benches showing us that that low of a clock speed using the Zen2 architecture you aren't coming close to a 10th gen i5, in pure gaming you might be looking at i3 level performance- that is best case for the consoles, the PS5 should probably limit itself to comparisons with laptops. A 10600k smashes a 3600(non X) which is better than a series X can hope to achieve.
Sony's NVME tech does indeed look nice, and I'll give Sweeney/Cerney credit on using up a cores worth of performance, I'll give them that, but most games have me showing a dozen threads using less than 10% load at any point- there is a reason Intel still has the best gaming CPUs, and it isn't "moar corez'. The consoles have been relying on heavy threading for many years now, one core more or less pales next to single core throughout for games.
Furthermore, both Sony and Microsoft dedicate cores to OS overhead, the dedicated hardware compression only in essence gives back what the base platform takes away.
On the GPU side, if it was 2018 you'd have a bit of an exaggerated point, but it isn't. Ampere is going to be here first with a full node drop, at best the consoles will be in the 3060 range of performance, but most likely lower still.
RDNA 2 reportedly has a 50% performance per watt edge over RDNA, quick math shows us this is going to put any APU *far* behind ampere, the series X is going to have less than 250 watts(likely far less) to play with, the PS5 even lower.
To build a PC that couldn't stomp the new consoles for $1k in November of 2020 you'd be a failure of a builder.
With all of that said, what you should be focusing on is the staggering improvement it is over that utter trash we had this generation. The PS3 and 360 threw down, they had zone impressive internals, this generation got their CPUs because they were too weak to put into tablets when they came out, that's no joke, Apple's A series and the Tegra CPUs smoked jaguar, they were utter trash. At least this generation we are getting something that can reasonably be compared to desktop CPUs, albeit much lower on the totem poll than what you believe.
Sony's loading tech is legit impressive, and one area where they will have a legit edge over PCs, I'll enjoy that when I fire up GT7 to be sure, but in no stretch of the imagination are the upcoming consoles going to be close to a high end gaming PC, it just ain't happening.
Owner of XBone, multiple Switch and PS4 and 4Pro consoles, outside of the bone being a waste of space I've been happy gaming on my consoles this generation, but they have been seriously weak from day one compared to desktop. This gen will be better, but not close.
First off you are absurdly off on your CPU projections, in ways that don't even make sense.
Best case scenario for upcoming consoles is the Xbox when talking power and that at best hits 3.8GHZ, we have several thousand benches showing us that that low of a clock speed using the Zen2 architecture you aren't coming close to a 10th gen i5, in pure gaming you might be looking at i3 level performance- that is best case for the consoles, the PS5 should probably limit itself to comparisons with laptops. A 10600k smashes a 3600(non X) which is better than a series X can hope to achieve.
Sony's NVME tech does indeed look nice, and I'll give Sweeney/Cerney credit on using up a cores worth of performance, I'll give them that, but most games have me showing a dozen threads using less than 10% load at any point- there is a reason Intel still has the best gaming CPUs, and it isn't "moar corez'. The consoles have been relying on heavy threading for many years now, one core more or less pales next to single core throughout for games.
Furthermore, both Sony and Microsoft dedicate cores to OS overhead, the dedicated hardware compression only in essence gives back what the base platform takes away.
On the GPU side, if it was 2018 you'd have a bit of an exaggerated point, but it isn't. Ampere is going to be here first with a full node drop, at best the consoles will be in the 3060 range of performance, but most likely lower still.
RDNA 2 reportedly has a 50% performance per watt edge over RDNA, quick math shows us this is going to put any APU *far* behind ampere, the series X is going to have less than 250 watts(likely far less) to play with, the PS5 even lower.
To build a PC that couldn't stomp the new consoles for $1k in November of 2020 you'd be a failure of a builder.
With all of that said, what you should be focusing on is the staggering improvement it is over that utter trash we had this generation. The PS3 and 360 threw down, they had zone impressive internals, this generation got their CPUs because they were too weak to put into tablets when they came out, that's no joke, Apple's A series and the Tegra CPUs smoked jaguar, they were utter trash. At least this generation we are getting something that can reasonably be compared to desktop CPUs, albeit much lower on the totem poll than what you believe.
Sony's loading tech is legit impressive, and one area where they will have a legit edge over PCs, I'll enjoy that when I fire up GT7 to be sure, but in no stretch of the imagination are the upcoming consoles going to be close to a high end gaming PC, it just ain't happening.
Owner of XBone, multiple Switch and PS4 and 4Pro consoles, outside of the bone being a waste of space I've been happy gaming on my consoles this generation, but they have been seriously weak from day one compared to desktop. This gen will be better, but not close.
SteveGrabowski
Diamond Member
- Oct 20, 2014
- 8,575
- 7,196
- 136
purbeast0
No Lifer
- Sep 13, 2001
- 53,455
- 6,301
- 126
It is definitely cheap. I wouldn't call it even close to awesome though, hence overrated. It is decent. And it has the worst fries in the business.
Double double is nearly $5 by me.
Straight from Anandtech:
"The CPU time saved by these decompression units sounds astounding: the equivalent of about 9 Zen 2 CPU cores for the PS5, and about 5 for the Xbox Series X. Keep in mind these are peak numbers that assume the SSD bandwidth is being fully utilized—real games won't be able to keep these SSDs 100% busy, so they wouldn't need quite so much CPU power for decompression."
It seems like the decompression chip inside the PS5 that removes high speed SSD bottlenecks is worth a few Zen2 cores itself. Of course, PCs won't come with dedicated decompression chips so the CPU will have to brute force the work.
Add in console optimizations, and a dedicated audio chip, a 3900x is not farfetched.
This is false. On consoles, games have maxed out the 8 cores of the Jaguar chip. It's just that on PCs, 8 core systems haven't been common so developers don't optimize for it. And for cross-platform games, a 4-core Intel chip easily killed an 8-core Jaguar chip.
PCs also have OS overhead. Moot point.
I'm referring to hardware available now: What would it take now to build a PC that can play games at the quality of a PS5/XSX. We don't know how future hardware will perform.
Last edited:
During a level load sequence a console will have lower CPU overhead somehow equates to it being on par with a 3900x......?
We have thousands of gaming benchmarks, why don't you look at how much of a difference there is in frame rate between a Gen4 Nvme and a HDD. This discussion is bizarre and absurd. The PS5 will load content very quickly, other then that its CPU is, being very kind, mediocre at best.
The cores aren't all maxed on a console, the system has at least one(if not two) locked for OS use, games aren't allowed to touch them. This is by design so system functions remain responsive. You can argue Windows *should* do this, but it doesn't. Next, the cores are pegged on a jaguar CPU because they are very, very weak. Take cross platform games and compare them in CPU limited situations, an ancient i3 beats down the consoles.
My OS overhead is normally single digit percentage on one thread, fifteen of them untouched.
If you want to talk about hardware available now, as you explicitly stated, then you can build a PC that smokes the PS4 Pro or the Xbox one X for a whole lot less them a grand. We don't know how future hardware will perform? Actually, we have a pretty good idea with the available information, you just seem to lack the level of knowledge on the PC side that you have on the console side.
We have thousands of gaming benchmarks, why don't you look at how much of a difference there is in frame rate between a Gen4 Nvme and a HDD. This discussion is bizarre and absurd. The PS5 will load content very quickly, other then that its CPU is, being very kind, mediocre at best.
The cores aren't all maxed on a console, the system has at least one(if not two) locked for OS use, games aren't allowed to touch them. This is by design so system functions remain responsive. You can argue Windows *should* do this, but it doesn't. Next, the cores are pegged on a jaguar CPU because they are very, very weak. Take cross platform games and compare them in CPU limited situations, an ancient i3 beats down the consoles.
My OS overhead is normally single digit percentage on one thread, fifteen of them untouched.
If you want to talk about hardware available now, as you explicitly stated, then you can build a PC that smokes the PS4 Pro or the Xbox one X for a whole lot less them a grand. We don't know how future hardware will perform? Actually, we have a pretty good idea with the available information, you just seem to lack the level of knowledge on the PC side that you have on the console side.
In addition, it's not just a single load sequence. Future games with SSD optimization will simply stream assets - hence, the need for a dedicated decompression chip.
First, I never said a PCIE4 (no such thing as Gen4 NVME) SSD increases framerates.
However, it seems to indirectly boost processing and graphically fidelity.
With an ultra-fast SSD, games don't have to pre-load as much into RAM. They can stream more data into RAM instead of preloading everything. This means an increased in fidelity and less RAM is needed to achieve similar effects.
Second, the dedicated decompression chip offloads data fetching from the CPU. According to Anandtech, this chip seems to be equivalent to 9 Zen2 cores on the PS5 and 5 Zen2 cores on the XSX at decompression.
Perhaps if you remove the marketing hype, it's 4 cores for the PS5 and 2 for the XSX. Still no slouch.
If you own an AMD 3600x, and games dedicate 2 cores to data decompression, well now you only have 4 total cores to work with.
Easy math.
It's an 8-core Zen2 CPU. It's not mediocre. It's certainly significantly faster than the average PC install base and certainly faster than the 3600x which is considered as mid-range currently.
In addition to the Zen2 CPU, it has a dedicated decompression chip (equivalent to 9 Zen2 cores at decompression according to Anandtech), and a dedicated audio chip (equivalent in audio power to the entire 8-core PS4 CPU according to Mark Cerny).
Now add in console optimizations, and a 3900x is minimum to equate the computing power to match next-gen console games.
I don't understand what your point is here.
Yes, consoles reserve one core for the OS. This is done for the responsiveness of the OS as you mentioned but also as a way to guarantee developers a certain amount of CPU power.
On Windows, even though a core isn't dedicated to the OS, Windows will allocate cores when there's an OS or other app task even while you're gaming. This can sometimes cause hiccups in games.
Just because Windows doesn't dedicate at least one core to the OS/other apps while you're gaming doesn't mean that games can always utilize all cores at all times.
Very good.
I'm not a CPU designer or a hardware engineer. I'm merely a software engineer. However, I do follow hardware news closely because I own a lot of AMD, Nvidia, Qualcomm, and Apple stocks. So maybe I do lack the knowledge.
Last edited:
Uh, your comments seem very detached from each other.
Follow the market very closely? I could provide hundreds if not thousands of links on this point, I didn't create the terminology.
Also, you seem.... confused as to how any real world comparison is going to work between the consoles and PC. The PS5 is going to load faster, and the PC will dedicate zero CPU time towards this fixed function task. If you have ever studied any element of comp sci you will understand that they would be an idiot to emulate a dedicated hardware solution from a dedicated platform to an open platform with no support whatsoever.
So the PS5 will load slightly faster and then be far behind in any CPU bound situation. This is going to be the reality. You can dream of new development paradigms all you want, they'll be limited to PS5 exclusives, and I'm sure they'll show us something useful and interesting not that awful R&C gimmick they have cooked up now(reminds me of Blinx(?) On the original Xbox).
Furthermore, assuming AMD's hype is real, 50% performance per watt improvement over RDNA, simple thermal limitations from the APU form factor tell us this isn't going to be close to the level of desktop GPUs, that is if AMD isn't overstating at all.
So in your mind, an 8 core 3.5GHZ CPU packaged as an APU with hard thermal limits is going to be comparable to a 12 core @3.9 GHZ or even an Oct core at 4.4GHZ?
It's got a cool sound chip and a cool chip to speed up load times, meanwhile the PC will be running at higher frames while pushing more pixels. If you think that makes it better, all the power to you, it didn't make it as powerful.
Follow the market very closely? I could provide hundreds if not thousands of links on this point, I didn't create the terminology.
Also, you seem.... confused as to how any real world comparison is going to work between the consoles and PC. The PS5 is going to load faster, and the PC will dedicate zero CPU time towards this fixed function task. If you have ever studied any element of comp sci you will understand that they would be an idiot to emulate a dedicated hardware solution from a dedicated platform to an open platform with no support whatsoever.
So the PS5 will load slightly faster and then be far behind in any CPU bound situation. This is going to be the reality. You can dream of new development paradigms all you want, they'll be limited to PS5 exclusives, and I'm sure they'll show us something useful and interesting not that awful R&C gimmick they have cooked up now(reminds me of Blinx(?) On the original Xbox).
Furthermore, assuming AMD's hype is real, 50% performance per watt improvement over RDNA, simple thermal limitations from the APU form factor tell us this isn't going to be close to the level of desktop GPUs, that is if AMD isn't overstating at all.
So in your mind, an 8 core 3.5GHZ CPU packaged as an APU with hard thermal limits is going to be comparable to a 12 core @3.9 GHZ or even an Oct core at 4.4GHZ?
It's got a cool sound chip and a cool chip to speed up load times, meanwhile the PC will be running at higher frames while pushing more pixels. If you think that makes it better, all the power to you, it didn't make it as powerful.
cmdrdredd
Lifer
- Dec 12, 2001
- 27,052
- 357
- 126
That sucks I guess. Not too worried, but I do feel like they should strive to get 60fps more too.
This is so hokey, I would love to know how they came up with this number. GPUs already decompress textures, is more needed?
They are compressing all of the data on the NVME, it will reduce install size make loading faster the way they are doing it. It legit looks like very nice tech.
Games built according to Sony's next-gen visions won't have level loading sequences. They could - and probably will - stream large quantities of data at any time. Level loading sequences exist because storage devices have traditionally been fairly to very slow compared to the amount of data loaded. If your storage device is instead very fast you could haul in any data you need just before you actually need it, continuously.
PS5 will be way faster than your average gaming PC. Take a look at the Steam survey and cry at what most people out there is gaming on. Mediocre is decidedly not the right word!
The next-gen consoles are actually quite potent, something you couldn't really say about the current PS4 generation.
Hmmh, hokey, sure. Quite possible. However, texture compression is designed to be simple, fast and allow (more or less) random access to the texture. If you're somewhat familiar to the way compression works, you probably know the basic principle is you search a batch of data for similarities and throw out the redundant bits.
The bigger the batches you search for similarities, the more efficient your compression, but the slower and more cumbersome it is to decompress. So for texture compression, the way GPUs typically use textures it means you can't use very big batches or it'd be no point in compressing the textures in the first place - the point is to reduce the amount of data not just stored in GPU memory, but also reduce the amount loaded from memory into the GPU.
Often you don't actually need all the texels out of a texture, so traditional texture compression uses quite small batches of texels when compressing - the original S3TC from way back worked with grids of just 2 by 2 texels as I recall. So compression ratio will be fairly limited. If you had textures compressed into big batches of say 16 by 16 texels or maybe more, you would have to load and decompress all of that data even if all you needed was a single texel somewhere in the map. It could easily end up being terribly inefficient (and in today's world, computing resource heavy, and thus power hungry, hot-running and clock throttling...
)So it would absolutely be beneficient with a higher order, much more heavy duty compression scheme for storing the game on the console's SSD before being loaded into RAM.
Yes, but a $1k gaming setup built right now is way faster than an average gaming PC, what I take issue with is needing $2k to build something comparable to a PS5 when it is available, that's laughable.
I disputed that myself earlier in this thread, and the OP chose not to engage.
You may draw your own conclusions...
GodisanAtheist
Diamond Member
- Nov 16, 2006
- 7,888
- 8,982
- 136
I think its hard to argue that this console cycle is definitely going to narrow the distance between current gen PC hardware and console hardware more significantly than virtually any prior release has done.
The real X factor here is going to be 6 months to 1 year out from the new console launches, what the price performance landscape looks like with Ampere and RDNA2 gpus, Zen 3, DDR5, PCIe 4.0 NVME drives etc hitting the market.
The real X factor here is going to be 6 months to 1 year out from the new console launches, what the price performance landscape looks like with Ampere and RDNA2 gpus, Zen 3, DDR5, PCIe 4.0 NVME drives etc hitting the market.
It's better than last gen on a relative basis, but it's quite a bit weaker than the PS3/360 gen, doesn't really outshine the PS2/Xbox generation either, I guess you could make an argument it'll be closer than the PS1/N64 was at launch but that's really only if we look at the US launch dates and not Japan(the OG Voodoo hit two months before the N64 launched in the US.
Pretty much last gen was so terrible it makes this one look better which was easy to predict when the last gen launched.
Pretty much last gen was so terrible it makes this one look better which was easy to predict when the last gen launched.
GodisanAtheist
Diamond Member
- Nov 16, 2006
- 7,888
- 8,982
- 136
- The PS3 launched around the same time as the GTX 8800 destroyer of worlds. No way was that a close match...
The OG Xbox kinda used off the shelf components and the GPU was essentially somewhere in line with NV's GeForce 3 Ti200/Ti500.
The PS3 also had a CPU that obliterated desktop parts, the 360 hit a year prior to the 8800GTX. The OG XBox was comparable to the top tier of desktop GPUs for a year until the GF4 hit.
This gen is on the weak side compared to desktop PCs relative to previous generations.
This gen is on the weak side compared to desktop PCs relative to previous generations.
VirtualLarry
No Lifer
- Aug 25, 2001
- 56,570
- 10,202
- 126
LTT builds a "better than PS5 PC". Something like 10x the price of a PS5, he mentions. 13.5GB/sec NVMe PCI-E 4.0 x16 RAID SSDs on a 24-core ThreadRipper (3? 4? Whichever one has PCI-E 4.0)
Except of course it had unified memory, so everything in the xbox competed for memory bandwidth alongside the GPU, while the desktop PC add-in board had all that GPU memory entirely for itself.
That had quite a large performance impact I seem to recall devs saying back in the day.
Magic Carpet
Diamond Member
- Oct 2, 2011
- 3,477
- 233
- 106
When I tell people I built a gaming PC, you know what's *NEVER* been one of the top three questions?
What's your storage setup?
What's your storage setup?
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 24K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 19K
-
-
-
Facebook
Twitter