EPTComputer Build, Upgrade & Buying Resource Thread - Page 205
| Forum Index > Tech Support |
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. | ||
KwarK
United States44439 Posts
| ||
|
MisterFred
United States2033 Posts
 . Luckily your dream infected me. I didn't think I could fit a new fx-build into your budget. The lack of any need for graphics power made it work.Edit: and just to be clear 8 AMD cores isn't twice the power of 4 Intel cores. But for EVE Online at least, it should be close. Just not for, you know, pretty much anything else. | ||
|
Cyro
United Kingdom20345 Posts
Edit: and just to be clear 8 AMD cores isn't twice the power of 4 Intel cores. But for EVE Online at least, it should be close. Against Haswell? No, not even close. Faster than a comparably priced Intel CPU, though. For it to be twice as fast, it would have to be equally fast on every thread as the haswell cpu and also have 1:1 scaling, it has neither With full utilization, it would be competitive with i5 at stock v stock, or oc vs oc. I can't see it being a ton better though, just more affordable and not-inferior if you are doing stuff like this or video encoding On x264 for example: 1 piledriver module is worth about as much as one haswell core, or rather, 4-module is fighting with 4-core at high end - i7 pulls quite far ahead with HT Steamroller is big big IPC gain there, so intel needs HT to hold the one module to one core thing - like you see in ~3.4-3.6ghz i3 performing neck and neck with stock 7850k (4ghz turbo) video encoding is not exactly the same performance wise as a lot of games etc (haswell and steamroller seem to benefit a lot more in x264 than in many games) but 1 module = 1 core has always been much closer than 1 module = 2 cores (which would have to be true for 4-module piledriver to double performance of 4-core haswell) | ||
|
KwarK
United States44439 Posts
| ||
|
Cyro
United Kingdom20345 Posts
 Oh, i quoted it at the start. I was indeed :D It's easy to think 8=4*2 and that fx is great when utilized, but actual situation with 4module piledriver is more "competitive in performance, great in value when utilized.. terrible when not" IMO, and you can't always choose when it's not utilized which leads to me not liking it very much for a general gaming system with solid budget (but running a bunch of eve clients doesn't seem to be that, and budget is not exactly that of a 4670k OC + gtx760 system) | ||
|
Froadac
United States6733 Posts
Assuming I upgrade to an i5, I'll either get 650TI or 660, and then a 128gb ssd. Does that sound like a reasonable minimal bottleneck build? | ||
|
Ata
Canada356 Posts
On January 29 2014 11:13 skyR wrote: If it doesn't have to be modular than you could just get the Corsair CX430 for $20 after mail in rebate? alright, thx for the input. Guess I will do that. | ||
|
Cyro
United Kingdom20345 Posts
650ti = Kepler, 768 cudas, 64 TMU's, 16 ROP's, 1gb gddr5 VRAM on 128 bit bus 770 = Kepler, 1536 cudas, 128 TMU's, 32 ROP's, 2gb gddr5 VRAM on 256 bit bus considering both at similar clocks, 680/770 can pretty much be said to be exactly double one 650ti - is there anything else that's super relevant for performance? I'm not really aware of any other stuff - if you have twice as many cores at the same clock speed, twice as many TMU's, twice as many ROP's and twice as much memory bandwidth, it should scale pretty much 2x, right? It's hard to check numbers, because very few benches etc actually list the core clocks under load for kepler, and it varies by quite a bit card to card - mine says ~1100/1189 max boost on the box IIRC, yet runs at 1254 at stock and ~1280 at stock voltage 650ti is about my minimum for a discrete nvidia card, because there's a big gap between 650 and 650ti. i3 is still good - i5 having four core native helps, but i3 being clocked higher this gen than last helps it fight a lot more with hyperthreading etc. Because the logical threads from HT are threads 3 and 4 (as opposed to 5-6-7-8), they will be used by a lot more software (basically any software where you'd benefit from the 4 cores of i5) so that means the performance gain from i5 over i3 is likely to be like 1.6x. In a few niche situations, it'd be 2x - the types of applications that scale with cores, yet gain no performance with HT. The only thing i can name off the top of my head that does this are the chess simulation engines, so it's not a big worry. i3 to i5 is a good upgrade, but it may not be entirely mandatory. For streaming, it's nice. For a lot of stuff, more graphics muscle might be a preferable way of spending the money, but it depends game to game, and i5+good GPU is a nice baseline to have, rather than a more lopsided setup that might work better in many things, but worse in some Depending on how good dual graphics is on Kaveri when we have the drivers and such, 7850k+OC dual graphics with a lower end radeon card might be good value and fight with i3 (surpass in some areas, lose in others - which is good! it's better than losing everywhere imaginable, like fx8350 vs 4770k) while giving better performance/$ on graphics muscle, but there needs to be some improvement over last gen dual graphics for that, there should be but i'm waiting for more actual benches etc. I don't think the 7850k is a good buy right now if you're not using the iGPU; i also don't really see the niche for 1080p gaming on entirely the iGPU either, but it seems to make some sense if dual graphics works good; at the very least, it's unlocked, which makes it worthwhile because it seems you can clock them reasonably, like 10% min on CPU and i've seen simultaneous with like +40% on the GPU, though it's probably not scaling performance with clock ideally (due to GPU being starved for memory bandwidth etc). Mantle helping dual graphics and asymmetric crossfire, maybe a real notable performance gain in mantle-enabled GPU bound cases vs nvidia having shadowplay etc actually makes kaveri+dualgraphics vs i3+~650ti quite different - I don't believe a lot of the high numbers for mantle performance apply to realistic gpu bound situations though, so i'm holding off on if it be a massive weight in performance/dollar on radeon vs geforce GPU's Mantle performance preview in + Show Spoiler + December January I accidentally a lot of words and didn't even get out caffeine yet~  | ||
|
Incognoto
France10239 Posts
I read in a thread that the FX 6300 is also a great chip. You have to overlock it though. If you overclock it to 4.6+ GHz (which it seems this cheap is easily capable of doing), in well threaded applications it competes with an i5 at stock. So encoding and things like that. I also read in a thread that an overclocked FX 6300 will have equivalent single threaded performance as an i3 at stock. I'm not entirely sure of this and the guy didn't back up his claim. He might have been talking about an IB i3 as the thread was old. Overall the preferred chips from AMD seemed to be the 83xx and the 6300. Most people who owned such chips seemed to be happy with them. There were a few exceptions. Also noteworthy is that a good motherboard and cooling is necessary to overclock even AMD processors. So even if the up front cost of the CPUs is cheaper, the overall system cost won't be that much cheaper, I think. | ||
|
Cyro
United Kingdom20345 Posts
I also read in a thread that an overclocked FX 6300 will have equivalent single threaded performance as an i3 at stock. It'd be close enough to i3 singlethread with that kind of OC, and stock i5 multi; i'd be a little concerned for sc2 performance as it seems to like cache/memory performance etc (and show huge preference to intel probably because of that) but for a lot of other stuff, maybe not so bad choice I don't actually know anybody on TL with an FX chip to bench against - i've seen some craazy numbers for this game, like Haswell doubling performance over FX even with a clock defecit. Would like to actually verify some - i'm not sure how valid sc2 performance numbers are anyway though (if anyone knows a better way for analyzing performance/fps, like FCAT for example but without a bunch of hardware, let me know*) + Show Spoiler + ![[image loading]](http://i.imgur.com/tva7K4s.png) ^Scrolling back and forth with game paused. Very smooth. ![[image loading]](http://i.imgur.com/pBesta2.png) ^UNSMOOTH (same thing that i did but with game unpaused), have no idea if fraps bench is at all accurate or how to quantify this, but it's really, really, really bad Oh and hey guys, i didn't realize that the NZXT H440 was so awesome. I put off looking at videos/reviews of it - i should have at least taken a quick look. Native 3x120mm front, 3x120mm top with support for a pair of 360 rads, psu shroud, window etc - medium sized (not small enough to present any problems, not big enough to make it more difficult to direct and handle airflow *cough* Elysium *cough* Seems beautiful. Likely choice for a new case for me. | ||
|
felisconcolori
United States6169 Posts
On January 29 2014 16:18 Cyro wrote: It'd be close enough to i3 singlethread with that kind of OC, and stock i5 multi; i'd be a little concerned for sc2 performance as it seems to like cache/memory performance etc (and show huge preference to intel probably because of that) but for a lot of other stuff, maybe not so bad choice I don't actually know anybody on TL with an FX chip to bench against - i've seen some craazy numbers for this game, like Haswell doubling performance over FX even with a clock defecit. Would like to actually verify some - i'm not sure how valid sc2 performance numbers are anyway though (if anyone knows a better way for analyzing performance/fps, like FCAT for example but without a bunch of hardware, let me know*) + Show Spoiler + ^Scrolling back and forth with game paused. Very smooth. ^UNSMOOTH (same thing that i did but with game unpaused), have no idea if fraps bench is at all accurate or how to quantify this, but it's really, really, really bad Oh and hey guys, i didn't realize that the NZXT H440 was so awesome. I put off looking at videos/reviews of it - i should have at least taken a quick look. Native 3x120mm front, 3x120mm top with support for a pair of 360 rads, psu shroud, window etc - medium sized (not small enough to present any problems, not big enough to make it more difficult to direct and handle airflow *cough* Elysium *cough* Seems beautiful. Likely choice for a new case for me. Hmm. Is a PSU shroud that important? In my Arc Midi R2 case, the PSU airflow is from under the case through the PSU and out the back; no air ever goes from the PSU into the rest of the case. | ||
|
Ropid
Germany3557 Posts
 | ||
|
Cyro
United Kingdom20345 Posts
+ Show Spoiler + ![[image loading]](http://i.imgur.com/qpjveEvl.jpg) ![[image loading]](http://i.imgur.com/NkewkTgl.jpg) ^If i have this.. I wanna see it ;3 | ||
|
felisconcolori
United States6169 Posts
On January 29 2014 18:44 Cyro wrote: Yea it's just kinda cool to have a lot of stuff that's in the case + Show Spoiler + ^If i have this.. I wanna see it ;3 Huh. And I thought the Noctua you all recommended was a big heatsink. | ||
|
Incognoto
France10239 Posts
| ||
|
Cyro
United Kingdom20345 Posts
On January 29 2014 19:18 Incognoto wrote: The "NZXT." is a tad ugly though. :p Sorry you have a typo there the word you are looking for is "sexy" it matches color scheme with ty147's ;w; Oh you mean the LED, not the case? I don't mind a white LED on a black/white design On January 29 2014 19:14 felisconcolori wrote: Huh. And I thought the Noctua you all recommended was a big heatsink. Need something to handle dat 200w haswell quad core (1.4vcore ht on) or fx (way more at 1.5v) or sb-e/ib-e/hw-e Luckily in uk, that beauty cost less than 20% of what 4770k did;; | ||
Myrmidon
United States9452 Posts
The front door could stand to be less restrictive too. Some performance figures: http://www.hardwareluxx.de/index.php/artikel/hardware/gehaeuse/29397-nzxt-h440-im-test.html?start=5 http://www.hardware-360.com/nzxt-h440-mid-tower-case-review/5/ note: hardwareluxx just testing noise levels with stock fans and passive components inside the box. I guess you need three fans and not at that low rpm to get air into the case. On the other hand, the design would dampen noise of internal components better than others. Probably better for water cooling. | ||
|
Incognoto
France10239 Posts
Anyway, just discovered Team Viewer. It's pretty insane! :o | ||
|
Cyro
United Kingdom20345 Posts
If you have "only" a pair of 770's and a 4770k @1.3v, sure not much to worry about, only dealing with like 500w as opposed to 900, so you need way less rad space to make it quiet >and< cool. A single 770 and 4770k@1.3v? Not really a problem, you're not far over 300 watts at combined full load going into the water water cooling should be pretty silent. Functional water loops are very very quiet. Kickass ones are not ^.^ | ||
|
wptlzkwjd
Canada1240 Posts
| ||
| ||
