Should I get the GPU & PSU first or all the parts at the same time?

Marcus97

Reputable
May 26, 2014
42
0
4,530
Hi there, I have got a new job and basically within the next couple of months I plan on building a new PC, with a 120gb SSD, 2tb HDD, 12gb RAM, 750w PSU, i5-4690k & gtx 980. I was thinking about getting the gtx 980 and the power supply after the first month as I could use it with my current build, and then wait until the next month to get the rest of the parts as I would have the rest of the money needed then. Would it be worth getting the 980 and new PSU first to use with my current system for a month? Obviously there would be bottlenecks as my current CPU is the fx-6300, but I think it would be cool to be able to get to use the 980 a month before, or do you think I should just wait until after a couple of months to get all the parts at once?

Cheers.

(All the parts I plan to get if anybody is interested in seeing them: http://www.amazon.co.uk/registry/wishlist/DAIRY22ZZIZV)
 
Solution


No, because it's dual channel. You need to put the RAM in pairs. You would need four sticks of 4, not 3 sticks of four. You are breaking the dual channel configuration and forcing the chip to run in single channel, which will reduce performance.
12 GB of RAM doesn't make sense for a dual channel memory controller.
You would want 8 or 16 GB.

That chip probably won't start to bottleneck until you start running SLI 980's. You might get lower performance in some games, but not the majority.
 


I just went with the extra 4 gigs for a bit of future proofing tbh. My computer at idle already uses around 30% of the 8 gigs of ram I already have so I thought it would help. Would it not work with the motherboard if I got those 12 gigs?
 

I would buy the rest of the parts after a month or so, the SSD, a new copy of windows 10 and everything else so I don't think that should be a problem, let me know if I am wrong though.
 


No, you need to pair the RAM sticks on a dual channel board.
2x4
2x8
2x8
2 sticks at those values. 12 GB isn't a matching pair, it's not a good configuration for a dual channel motherboard.

When you use DDR3 RAM sticks, they need to be the same brand, same speed, and same size preferably. Any other combination will create performance issues and instability.
 
I see no problem with buying the graphics card and psu now. You get to use them and know that they work.

On the rest of your build, 12gb is a strange number making 4gb run in single channel mode.
Much better to buy a single matched 16gb kit of 2 x 8gb.
There is no guarantee that sticks from separate kits will work together.

 


Oh ok, I read that if where to get the 12 gigs the 2 X 4 gigs would run in dual channel mode and the other 4 gigs would run in single channel mode, is that right? Would there be a noticeable decrease in performance than if I where to just have two sticks?
 


I made sure that that extra 4 gigs is the same make/ model as the 2 x 4 gigs, would that be ok?

 


No, because it's dual channel. You need to put the RAM in pairs. You would need four sticks of 4, not 3 sticks of four. You are breaking the dual channel configuration and forcing the chip to run in single channel, which will reduce performance.
 
Solution


Ah ok, I understand now, thanks. Would the performance reduction be noticable, say if I where in game would there be an FPS drop or something?
 


https://www.youtube.com/watch?v=-D8fhsXqq4o

Watch, he explains it well.

Performance? Yes. Maybe not noticeable in games.

Stability? Absolutely.
 


You already seem to finish the topic with choosing the best answer... let me just say this:
Look over DX 12 stuff and you are about to realize the whole new buy might not be so shiny after all.
At you position I'd rather go for FX8350 if possible with your mobo and save the cash for later.
Another thing is GTX 980.. also in this case i would simply skip few month until Fury settles in price and go for it.
12 GB doesn't make any sense as described, get 2x4 GB of non exotic memory so in case somehow it lacks you can get similar grade another 2x4GB anytime later.
When you read about DX12 you will realize how underestimated AMD was before with DX11.
Now to be clear that i5 and GTX 980 are excellent choices for the time being considering it will take time for devs to actually make a decent DX12 games but once that gonna happen:
1 - AMD CPU's will get a fresh breath in gaming due lower CPU overhead boosted with higher core count.
2 - nVidia will look like a infant in a sandbox comparing to AMD GPU's due to parallelization and computation power.

People are finally starting to get what just happened here with Intel, nVidia and M$.
AMD was a drudge for over a decade f..ed by those 3 above, they fact that they didn't bankrupt yet is simply admirable, the fact that to survive they need to fight both with prices and new ideas is even more admirable, but the fact they are so close from flipping the coin in this conditions is mind blowing.

Now to be clear, I ain't no AMD fanboy, their CPU's do suck in single threaded performance which is still important because of lazy devs and ridiculous DX11 implementation which favors fast cores, AMD GPU's loose against nVidia's counterparts because DX11 can't take full advantage of their horsepower but next year seems to be quite surprising from what i read:
http://www.overclock.net/t/1569897/various-ashes-of-the-singularity-dx12-benchmarks

 
I would disagree with all of the Above. Intel CPU's outperform AMD's, and are about 4 years more advanced at this point. The comparison is done, only the price to performance ratio exists.

A single computation requires a single thread, it can't be divided over multiple threads. So the performance of single thread will always be important. You can't program around that requirement, which is why intel i3's can blow ahead.

As for AMD GPU's, they've been saying that about performance for years. "After this update.. after windows 8, after this after that.."
It just doesn't add up in the real world benchmarks and they always lag behind. I wouldn't invest on empty promises.
 


This either mean i was wrong in every single word or you are a hypocrite.
Please be more specific about me being wrong, because up to this point it looks like simply trolling.

"Intel CPU's outperform AMD's, and are about 4 years more advanced at this point."
First part is a generalization i was pointing to swap to i5 platform as not necessarily so shiny as the OP might thing.
Second part of the sentence ... whatever:
http://www.techspot.com/community/topics/then-and-now-a-decade-of-intel-cpus-compared-from-conroe-to-haswell.217443/

"As for AMD GPU's, they've been saying that about performance for years. "After this update.. after windows 8, after this after that..It just doesn't add up in the real world benchmarks and they always lag behind. I wouldn't invest on empty promises. "

Not sure what is your conclusion from this sentence.
I seriously don't care which side is at the top of the chart as far as i get worthy piece of technology and imo AMD cards are victims of their own complexity. From few certain families they are designed to move the boundaries but the whole game ecosystem is in stagnation and laziness. There is certainly far more potential on those cards then people see with "real world benchmarks". Those real world benchmarks are based on game engines that are full of flaws operating on a API that is worth a shit. Its not in market interest to take out the best of both hardware and software, its meant to rip off people from cash for the illusion of technology progression. It seems Intel and nVidia are very welcome about this situation indeed.
DX11 and the M$ aversion to make a decent API is an excellent evidence of that.
Read more about DX12, Mantle and Vulcan you will realize whom we should thank for it.



Hardly possible for me to be hurt since I'm not the one. I'm open into discussion with fanboys tho i might learn something i didn't know or didn't realize 😉
 
guys make your own thread to debate this... it really has no relevance for the initial question. he knows already what he wants to buy, just doesnt know / can't decide if it's ok to get them separate.

i'm not taking any sides here, just... "take it outside"
 

In my opinion, your opinion is completely based on brand preference.
 


Link shows nothing, the AMD chips still run 32 nm, use significantly more power, run to hot, and even AMD avoids their own CPU's in their own towers.

Eerything AMD has released in the last 2-3 years has been a major disappointment.

Regardless of who blame, they don't keep up. By the time DX12 is standardized, Nvidia will have already released a new product and pulled way ahead again. Mantel is a low level API, and with the exception of low end videocards, it doesn't really improve much in the high end. I understand how low level API's work, they've been out for 2 decades, famously 3dfx Glide.

You seem to think that magic API's and drivers are going to make AMD's CPU's and GPU's perform better. By the time they release a driver, Nvidia will release new technology. They are too slow to adapt. I wouldn't buy an AMD GPU that does not perform as well as Nvidia, does not get the same level of driver releases and requires a liquid cooler to keep stable on the false-promise that an API that will be very slowly adapted will increase perform marginally.

" Unfortunately, HBM is not the saving grace of the AMD Radeon R9 Fury X that propels it forward in 4K gaming currently. It is held back by capacity and performance. "
http://www.hardocp.com/article/2015/07/26/amd_radeon_r9_fury_x_4k_video_card_review/8#.Vf2jsPQYGUk

I can't compare CPU's, because AMD doesn't make a CPU that really competes. You sound like a conspiracy theorist after that post. AMD cards are no more or less complicated than Nvidia, the company just can't get the performance they want because they don't have the capital to invest in R&D.

tl;dr
OP, the new line of AMD cards are just rebranded 2XX series, with the exception of the Fury. For the price, the Nvidia card is a better buy. If magic drivers and API upgrades make the Fury faster, it will be maybe 5-10% in the odd title and Nvidia will release a new driver by then.

As for the CPU, the FX-6300 isn't a major bottleneck for your videocard, but you will still get increased performance on a newer Intel chip. You can blame the programmers and wait for better programming over the next decade, and your 10 years old processor might improve by than. But just accepting the fact that a single thread in Intel can run two tasks as fast as 2 threads can run 1 in AMD, is reason enough not too.
Again, by time programs become better as parallel, chips will come with new instruction sets and it would be pointless. This guys arguments are not 6 months from now, they'll be applicable in a few years, when all current GPU's and CPU's will have been refreshed anyway.

This isn't a fan boy argument. I have 2 systems, both running AMD chips. But my gaming rig is 3 years old, and I won't deny that when the time comes to replace it, Intel is the only option on the market.
 


Thats a nice generalization, I bet those who bought their CPU's and GPU's where quite satisfied for the money spent.

Unfortunately calling Mantle, Vulcan or DX12 a magic API you crossed the point when I hardly want to be involved in further discussion.

For those who are still open-minded let me again suggest a nice read, i know it's long but it puts a nice light on the current situation:
http://www.overclock.net/t/1569897/various-ashes-of-the-singularity-dx12-benchmarks
 


When actual DX12 titles are released, in 1-2 years from now, than we can start talking after Nvidia releases it's new cards, or new drivers.

Basically, for the same price as Nvidia, you can have sub-par DX11 performance, or equal DX12 performance with fewer driver updates.