https://www.digitaltrends.com/computing/amd-ryzen-5900x-vs-i...
Is $100 really that much? Get a second hand gpu from 5 years ago for less. Those don't have issue with video though you might have to look for right connectors.
a geforce 1030 will be adequately taxed on desktop loads: https://www.techpowerup.com/gpu-specs/geforce-gt-1030.c2954
Can get them for 40eur, they barely do 1080p.
You can go DVI to HDMI with either a 5$ adapter or a single DVI to HDMI cable. Only real difference is DVI doesn’t carry sound, but as long as one monitor has sound that’s irrelevant.
"Number of DVI inputs 0" ( https://www.bestbuy.com/site/samsung-ur55-series-28-ips-4k-u... )
And the GT 1030 is only single-link DVI, not dual-link, so you can't even drive 2560x1600 with it.
So it tops out at 2x 1080p monitors (or 1 high-res + 1 1080p secondary, whatever). Literally worse than GPUs from 10 years ago. Which since all I want a GT 1030 for is to drive monitors makes no sense at all.
But, again the cheapest dual 4K/60fps video card costs more than buying 2 of the cheapest HDMI 2.0+ cards. Which is why graph is card companies don’t really cater to that market. Unless your case and motherboard only has slots for a single card it’s effectively a non issue.
I still don’t hear an argument here for having it all in one package. There are APUs in the Ryzen series, they perform worse, but everything is a trade off, you can’t have your cake and eat it.
You can get new and adequate GPUs for less than 100eur which will do what you want and “won’t sit idle” because they’re taxed. This GPU would struggle with even 1 screen. I think you underestimate the load a desktop has on a GPU, 2D rendering isn’t “free” and can take a lot of CPU die space and add to the heat/power delivery issues of a modern cpu.
From what I've seen, the APUs perform pretty similar to the CPUs they're derived from. The thing is the APUs are the last thing AMD releases, and the model numbers are confusing, and Zen2 APUs aren't widely available. Zen+ quad core isn't very exciting anymore, but it's the APU you can buy at retail in the US.
- No need for a separate component, no hassle finding a "proper" card and buying it (I.e. you only need basically a integrated GPU as a PCI card, but good look getting one, you always can get older "cheap" GPUs but then there is always the chance that because them being older the driver support started to rot.)
- Integrated GPUs tend to have reasonable good Linux driver support, at least for newish CPUs for a view years.
- No extra fan or other components which can make unwanted noise.
- One less component makes trouble shooting easier.
Tbh. if AMD would ship something like a large package (not on-die, but chiplets or otherwise packed) combining a 16+ core cpu (+hyperthreading) with a reasonable integrate GPU and ~2GB low latency RAM pro core + 3year waranty I would be quite interested in that. Some companies I worked with before would like that too as far as I can tell.
- developer machine (I don't game)
- server in the basement
- htpc, no gaming (maybe some retro)
In all of these (and probably more), a discrete GPU feels like a detriment. More space, more heat, more drivers, more hassle, more money.
Ofcourse there are related subdomains where I need the graphics, but we are talking about 1 in 1000 use case.
I just need the gfx card to drive the display only.
But then you need another graphics card for your main Linux system.
I see the use case at least, but I agree that I rather AMD use the die space for a faster CPU and release cheap (sub $100) and low power graphics cards with multiple display outputs.
They don't have to use die space for it. CPU cores are in a separate die already. They can easily slap a Vega 3 IC (which they already have the designs for) alongside the 12nm I/O die.
> release cheap (sub $100) and low power graphics cards with multiple display outputs.
The problem is they're not doing that, either.
Last time they released a "low" end GPU was 5 years ago with RX 550. And it wasn't actually low-end, price-wise (It had $80-100 MSRP and never got cheaper thanks to crypto miners)
These CPUs are using a multi chip design. There'd be no waste of die space. They can slap a small Vega3@12nm die alongside I/O die.
- limited production capabilities
- maybe not-so-perfect yield
- self-competition to their still selling 3000 processors and their son to (theoretically) sell discrete graphics cards
To get a chance to chip Ryzen laptops AMD needs to chip a APUs to OEMs similar for the office pre-build desktop marked. So chipping APUs to them is necessary, on the other-hand shipping it to end users is not necessary and in context of limited production capabilities it's a bad idea to ship an additional product to a marked where (in difference to OEMs) it's hard to judge how many will be sold beforehand and where selling them can negatively affect selling other AMD products.
They have a thing going where they manufacture Zen core chiplets, and that put together X number of chiplets to make a processor. This is very attractive in terms of cost.
As best I know they don't have a GPU chiplet, and so they can't just add one to a CPU. All their processors containing both CPU and GPU are designed like the M1 - one big chip with both.
Hopefully this will change soon, and we'll have Navi 2/3 chiplets going around.
That is, 5600X > 10600k, 5700X > 10700k, etc... This is, however, assuming MSRP pricing for both. If the Ryzen is being marked up due to shortage while the Intel is being discounted a significant enough price gap could warrant still going with Intel.
Gaming: https://www.pc-kombo.com/us/benchmark/games/cpu/compare?ids%...
Applications: https://www.pc-kombo.com/us/benchmark/apps/cpu/compare?ids[]... (that's a rather small selection though)
The 5800X is clearly faster.
It's the straightforward comparison, 8 cores/16 threads for both. Surprisingly, the 5600X is still faster - at least in games.