Would you buy a new gfx card for your 5 year old computer? I currently have a amd phenom 955 with 8 gigs of ram I bought 5 years ago and upgraded the vid card 3 years ago with a geforce 680 4 gig.
Currently thinking of building a new computer next year when I get working again.
Im looking at Nvidia geforce 980's something in the 6 gig range, but dont know what card is gud, theres a few on sale for anywhere $100-$150 off
tl;dr 5 year old computer + new vid card?
>>52062312
Now is a stupid time to get a video card. Upgrade that CPU first and wait for greenland/pascal.
>>52062360
Why a stupid time?
Upgrading cpu means new case power supply ram etc etc, as I'm gonna turn my old desktop into a media server
I dropped a 970 into a 920 system.
I don't play games on my computer, so my six year old computer's six year old video card is perfectly fine.
>>52062375
Next set of GPUs is coming in Q1 and it is going to be the first die shrink in 3 years/HBM2 so it should be a pretty big jump.
>>52062407
Yeah just like how skylake was a big jump too because it was a dieshrink
The age of dieshrinks causing performance improvements is over.
I'd love to be wrong here though, but I don't think I will be.
>>52062381
A OC'd l7 920 is still pretty decent.
A OC'd phenom 2 955 is not as decent by a long shot.
>>52062407
Any idea on the price ranges?
I'm not in a rush to buy or build anything yet, desktop still works ok for most games on medium /high
>>52062432
4.1 OC on the 920. It hurts a bit on single threaded games that are CPU intensive. But overall it hits more than 60fps on just about everything just about all the time at 1920x1200.
>>52062422
Current CPUs are not limited by their process. Current GPUs are. We are at the limit of 32nm hence all the housefires.
>>52062433
depends on how beefy you want it.
>>52062447
Right so the same way that CPUs get a massive bump in performance from 28nm to 14, GPUs will too.
>>52062463
I'm thinking my budget will be around 600 to 900ish CAD for the vid card alone, will probably go a bit higher for mobo end cpu eventually
>>52062478
Nigga do i need to repeat myself? GPUs designs are currently limited because it is not physically possible to make a bigger GPU on 32 nm without yields imploding to early Fermi levels. CPUs are not currently running into this issue at the consumer level. There was a massive increase in the amount of cores they could fit on one CPU from 28nm to 14nm. A process shrink does nothing without a architectural change. CPUs are not like GPUs. CPUs are monolithic. GPUs are HIGHLY parallel.
>>52062494
should be golden for a lower high-end card.
Since this thread is related
What's the best GPU you can get without bottlenecking an i5 4460? I'm using a 280x, I'm wondering if a 390 is already overkill.
>>52062543
nigger are you fucking retarded?
>Bottlenecks are myths
>>52062543
Problem is that die shrinks cause massive yield issues right now so the higher transistor counts might not end up happening for a long time.
http://pcpartpicker.com/p/VWJb7Ppretty set on the case I'm try not to spend more than 350 on this right now I'll be adding a 390 and an ssdin later how can I improve it?
>>52062622
http://pcpartpicker.com/p/VWJb7P sorry posting from my phone
>>52062554
Intel's consumer CPUs haven't changed much since sandybridge. Manufacturing prices have gone down and yields have gone up by on the consumer end it doesn't make a difference because the jobs consumer CPUs do are not highly parallel and therefore do not scale well across more than 4 cores. For those who actualy can use more than 4 cores effectivly there has been a significant jump in performance. We now have 18 core xeons something that was not practical on the old 28nm process. With the die shrink they will be releasing 22 core Xeons soon.
Now unlike consumer CPU workloads consumer GPU workloads are highly parallel. GPUs primarily improve by tossing MOAR CORES at the problem. We are at the point where there isn't die space to toss MOAR CORES at the problem without reaching into the negative profit margins thanks to poor yields.
http://renderingpipeline.com/2012/11/understanding-the-parallelism-of-gpus/
LRN 2 computer shitlord
