|
When using this resource, please read the opening post. The Tech Support forum regulars have helped create countless of desktop systems without any compensation. The least you can do is provide all of the information required for them to help you properly. |
United Kingdom20345 Posts
On March 15 2014 00:55 nosliw wrote:What is your budget?$800 USD What is your monitor's native resolution?1920 What games do you intend to play on this computer? What settings?SC2 max setting, multi player, and streaming What do you intend to use the computer for besides gaming?streaming SC2 Do you intend to overclock?No Do you intend to do SLI / Crossfire?No Do you need an operating system?No Do you need a monitor or any other peripherals and is this part of your budget?No If you have any requirements or brand preferences, please specify.Intel CPU, probably NVIDIA GPU because AMD cards are inflated. What country will you be buying your parts in?USA If you have any retailer preferences, please specify.None So I saw this part list from newegg It looks pretty similar to the template build in the OP. What do you guys think?
Depends really if you want to go cheap or powerful, also how much upload you have. i3 +750ti is really appealing there, but you might want i5 and 760 for other games.
Here's an example for something that would appeal for me if you want to shoot lower on the costs as opposed to big meaty system:
PCPartPicker part list / Price breakdown by merchant / Benchmarks
CPU: Intel Core i3-4130 3.4GHz Dual-Core Processor ($118.97 @ OutletPC)
Motherboard: MSI H81M-P33 Micro ATX LGA1150 Motherboard ($49.98 @ OutletPC)
Memory: Kingston Black 8GB (2 x 4GB) DDR3-1600 Memory ($64.99 @ Amazon)
Video Card: EVGA GeForce GTX 750 Ti 2GB Video Card ($159.99 @ NCIX US)
Power Supply: Corsair Builder 430W 80+ Bronze Certified ATX Power Supply ($39.99 @ Microcenter)
Total: $433.92
(Prices include shipping, taxes, and discounts when available.)
(Generated by PCPartPicker 2014-03-14 12:38 EDT-0400)
^Requires case, also requires some form of storage. You can throw in a 120gb 840 evo SSD for like $85, maybe you'd want that with a 1tb HDD, or just by itself, or a 250gb SSD by itself etc. You can also "obtain" an OS that you like. Such a build would also favor using the onboard encoder on the GPU (nvenc) for streaming, which gives better game performance than cpu encoding, requires less hardware investment, but has the cost of having less quality at a given bitrate, at least in certain circumstances
Oh, and not sure on the mobo. I don't really know the cheaper mobo's very well, somebody else will know important details though. I think that's the one that somebody else bought?
Pretty open ended (it's just a small collection of parts i guess) in that you could throw in an i5 ~4570 or ~4670 (or whatever is best for price to performance, sometimes one model is way cheaper) if you wanted to, or also open up to a stronger graphics card. A lot of stuff that would help other games won't really help sc2 performance, though
|
Ya I play other games too. What is a good balance between CPU and GPU? And I have issues understanding GTX series... Is there a logic to how they number the cards? how is 760 and 670 around the same mark?
|
On March 15 2014 02:04 nosliw wrote:
Ya I play other games too. What is a good balance between CPU and GPU? And I have issues understanding GTX series... Is there a logic to how they number the cards? how is 760 and 670 around the same mark?
GTX 400, 500, 600, 700, 800 are the generations.
GTX 460, 470, 480 is the price bracket.
If something mentions "Ti" at the end, its performance is halfway towards the next higher model, for example GTX 560 Ti is between GTX 560 and GTX 570.
So that GTX 670 is a higher price bracket than the GTX 760. GTX 760 is a newer generation and the improvements cause it to be equal or beat the GTX 670. Another example, the current GTX 760 is about twice as fast as the GTX 460 from three generations back.
|
United Kingdom20345 Posts
TBH i find it hard paying $100 more for 760 vs 750ti. The great 750ti's are clocking to like 1470 on 1.16v. 1350 and a bit of memory OC should come easy, they perform a bit too well for $160 vs $260 price bracket
|
Ah thanks, so GTX 750Ti is half generation above other GTX 700x cards. And the Ti has the Maxwell stuff?
|
Both GTX 750 and GTX 750 Ti are Maxwell.
|
ok, but GTX 760 is not maxwell? Just curious, is AMD card prices still inflated due to coin mining?
|
Yes, AMD prices are still inflated.
Only Maxwell cards out right now are the GTX 750 and GTX 750 Ti. Higher-end Maxwell will come out later in the year and will be the GTX 800 series.
|
On March 14 2014 11:19 skyR wrote:
Why that particular Kingston memory? It's 1600MHz at 1.65v aka bad. There are better performing and less expensive G.Skill and Corsair kits. All three companies sponsor esports fyi.
What speed/volt/CAS is a good combination? 1600@1.5V@CAS9? I'm looking for higher speed, lower volt, and lower CAS right?
|
On March 15 2014 02:46 nosliw wrote:
ok, but GTX 760 is not maxwell? Just curious, is AMD card prices still inflated due to coin mining?
Yeah, I forgot to mention those GTX 750 and GTX 750 Ti. Those two are really something out of the future GTX 800 line. NVIDIA probably didn't want to start with the 800 numbers as the replacements for the 760 and 770 and 780 are not yet done.
|
On March 15 2014 03:02 nosliw wrote:Show nested quote +On March 14 2014 11:19 skyR wrote:
Why that particular Kingston memory? It's 1600MHz at 1.65v aka bad. There are better performing and less expensive G.Skill and Corsair kits. All three companies sponsor esports fyi.
What speed/volt/CAS is a good combination? 1600@1.5V@CAS9? I'm looking for higher speed, lower volt, and lower CAS right?
Yes.
Your typical memory is 1600MHz cas9 @ 1.5v. Something better like 1866MHz @ 1.5v or 2133MHz @ 1.65v is not much more expensive in some cases.
|
United Kingdom20345 Posts
On March 15 2014 03:02 nosliw wrote:Show nested quote +On March 14 2014 11:19 skyR wrote:
Why that particular Kingston memory? It's 1600MHz at 1.65v aka bad. There are better performing and less expensive G.Skill and Corsair kits. All three companies sponsor esports fyi.
What speed/volt/CAS is a good combination? 1600@1.5V@CAS9? I'm looking for higher speed, lower volt, and lower CAS right?
Yes, but you can only use 1600mhz RAM unless you get an overclocking mobo which would cost more and not make sense unless you wanted to up budget a ton to overclock and cool a 4670k as well. 1.5v vs 1.65v doesn't matter so much; i just listed that because it was like $4 cheaper.
|
What settings/testing?
I had a pretty shitty CPU draw so I had to crank the volts up to 1.350 for 4.5GHz on all 4 cores. Did a prime and IBT for an hour each and didn't crash (maxed around 90 celsius). Played SC2 for a month or two before I left for my trip without crashing. I haven't tried OC'ing just 1 or 2 cores yet which might help with temps and be equally good for SC2.
Can't remember how much I OC'd my GPU right now but only by about 100-150Mhz since I had a 2GB Zotac GTX 770 OC. Just did some Heaven Unigine and Valley tests for that before getting into SC2.
|
4.5 @1.35 is only slightly worse than average. I'd play around with your VRIN to see if you can get that vcore down a bit.
I'd probably still run the custom looping x264 test from the OCN thread to get a longer running stress test in there. Which P95 are you using, and what test specifically?
|
United Kingdom20345 Posts
Echo everything mav said (though it's like 100mhz down from average if it can't be improved), everyone should read the OCN thread
Here's one of the more recent loops: http://download1346.mediafire.com/pf1y1esf6qhg/2laifu8xqi6vz2g/x264 Stability Test.7z
Example of running and my performance: + Show Spoiler +
Very relevant: ![[image loading]](http://cdn.overclock.net/3/36/500x1000px-LL-366c4dd8_z.png)
^High end air, i5, 1.25vcore set in bios, manual volts
There's little logic using IBT for example, as it's just older version of linpack that's not updated to use avx2 instructions. If you use the updated one, it's crazy hot and you have to use like 1.0-1.1vcore because of the massively increased power draw* - yet if you use regular IBT, it's just hotter than other tests without really being of more help to find the points where you are stable
*x264 pass 2 = ~116w
Peak Vcore = 1.080
Hottest core = 51c
LinX 0.6.5 with 6144 memory entered = ~170w (~192 GFlops)
Peak Vcore = 1.092
Hottest core = 78c
"It's at the wall for the entire system.
System:
4670k
2x4GB ram
7950
one five year old mechanical hard drive (was always spun up)
one SSD
two 120mm pwm fans
four 140mm fans
seasonic ss-660xp 600w psu (rated platinum)"
^1.05vcore set, entire system power draw peak 116w with full CPU load from encoding. Compared to Linpack with avx2.. power draw is 54w higher, and that's >added< to the power draw from full load CPU at ~1.08vcore. Wall numbers, but it's a seasonic plat rated PSU and the difference is very alarming.
Example here: If we take 20c case temp (which is optimistic even in a ~17c room) then the 51c result is 31c above ambient, the 78c result is 58c above ambient. Since the best way to measure temperatures is degrees over ambient and the difference measured by degrees C increases as you go higher on volts (shown above in chart) it's pretty insane
This is partly why it's common practice now to test for stability instead of worst case temperatures and you can do so quite effectively with things like x264, specific fft's set in some versions of prime and good overclocking practice
|
United Kingdom20345 Posts
Pulled cable out of third fan on windforce cooler, went from:
85c @~1000-1137mhz @80% fan while gaming
to
61c peak @1293mhz @80% fan after 5 minutes of heaven 4.0
gg
I will admit to being an idiot this time, had to route cables differently because GPU was too close to other stuff. I checked it once and assumed it was fine and i was quite badly wrong
![[image loading]](http://i.imgur.com/zMaaNuxl.png)
^This fan.
|
United States7481 Posts
you're saying it performed better without a fan running?
|
yeah i'm super confused as well
|
United Kingdom20345 Posts
went from:
85c @~1000-1137mhz @80% fan while gaming
to
61c peak @1293mhz @80% fan after 5 minutes of heaven 4.0
That's the before and after result for fixed fan. Hotter and slower result is the "broken"
|
I think he's saying there was a cable getting into the blades of that third fan and that fan was stuck because of it.
|
|
|
|
|
|
EPT![[image loading]](http://i.imgur.com/RI96ddS.png)
