Validate my build

[mockingbird]

Re: Validate my build

Yeah, they use a lot of power. FX-8350 is about 2.5x as much compared to the i5. But that's not a major concern. In my current dwelling I'm not billed for my electricity usage. (Just a flat rate.) It also has a higher benchmark score and costs £10 less, motherboards are cheaper too. Are there any good reasons, ignoring power consumption, for avoiding the AMD chip?

One word: Drivers

£10 is a very small price to pay for not having to put that atrocious AMD/ATI code on your system.

[tom66]

Re: Validate my build

I will admit I've always preferred Intel for Linux support too. Their Linux drivers are absolutely top notch. Then I saw this with some benchmarks:-
https://www.youtube.com/watch?v=JFqQ...U&noredirect=1

While ranking is higher, FPS is better on i5-4460 in all cases. So I'll probably stick with i5 in the end.

[cheapie]

Re: Validate my build

I'm running an AMD FX-8350 here, and it is (as said earlier) extremely fast if you can keep all 8 cores busy. For what I do, I can.

As far as drivers, I have a Radeon HD 7770 video card, which works perfectly fine with the open source "radeon" driver. I think it requires some proprietary firmware blobs, but at least it's not fglrx.

[tom66]

Re: Validate my build

Well, the two builds:

CPU: AMD FX-8350 4.0GHz 8-Core ... £125.00
Motherboard: Gigabyte GA-990FXA-UD3 ATX AM3+ ... £82.00
Memory: Kingston 16GB (2 x 8GB) DDR3-1333 ... £93.97
Power Supply: SeaSonic 650W ATX12V / EPS12V ... £87.99

Total £388.96

vs:

CPU: Intel Core i5-4460 3.2GHz Quad-Core ... £130.94
Motherboard: Gigabyte GA-Z87X-UD4H ATX LGA1150 ... £106.92
Memory: Kingston 16GB (2 x 8GB) DDR3-1333 ... £93.97
Power Supply: SeaSonic 650W ATX12V / EPS12V ... £87.99

Total £419.82

(I ended up changing the PSU, this one is not group regulated and looks better built, for only a little more £)

I'm still leaning Intel-wise, but will think carefully about what I need. A lot of CAD work runs only one or two threads so better single core execution may win it.

I'm still not convinced 1600MHz RAM makes much difference after seeing this:
http://i.imgur.com/V0NiI.jpg

A few % differences, at best...

[c_hegge]

Re: Validate my build

Definitely go with the Intel build. I would probably suggest going with a current Z97 board rather than a previous generation Z87 board. Bear in mind that a Z87 board will need a BIOS update in order to work with the i5 4460 (which might be a problem if you don't already have an older CPU like a 4440 to do it with).

EDIT http://www.cpu-world.com/Compare/368...5_i5-4460.html

According to CPU World, the 4460 and FX-8350 are both about the same performance in multi-threaded applications, but the i5 is noticeably faster with single threaded applications.

[tom66]

Re: Validate my build

Yeah, I don't have an older processor. So that would be an issue. Buying one is probably not worth it, unless I got the absolute minimum one, but probably still going to be ~£50. Would a Z97 support it out of the box?

[c_hegge]

Re: Validate my build

Yes. Z97 boards support Haswell refresh CPUs out of the box. I'm pretty sure they will also support Broadwell out of the box too. I would probably suggest something like a GA-Z97X-UD3H if you can get one.

[RJARRRPCGP]

Re: Validate my build

Yeah, they use a lot of power. FX-8350 is about 2.5x as much compared to the i5.

I doubt as much as my Q6600 @ 3.3!

A stock FX pwns my Q6600 @ 3.3 in video encoding.

A Kentsfield is a power hog, lol. And also, my Kentsfield apparently (or much more likely to) also strangely reports bus errors when I don't have sub-ambient temps @ 367 Mhz FSB.

Despite 1.3 V being nothing for FSB termination voltage on a 65nm...

I read that a Kentsfield @ 3.3 easily consumes 250 W at full load.

[Uranium-235]

Re: Validate my build

Mid-50s under full load with the Noctua cooler... With the stock cooler, I read reports of 80C+ under load, that's why I splurged on the cooler. $25 coolers are also fine, but it's not right that Intel is including these coolers on $200 CPUs. On Pentiums or even i3 I understand.

you read reports and didn't try it out yourself? Some people don't install coolers right

[mockingbird]

Re: Validate my build

you read reports and didn't try it out yourself? Some people don't install coolers right

Multiple reports... and they sounded credible to me.

But if the chip was sitting in the mid-50s at full load with a hefty Noctua cooler, it stands to reason that 80-90C would be the result with a 15 or so millimeter thick HS/F.

That machine was going to get a high-end cooler anyways. I had a Clarkdale system come in from the same customer with a pancake cooler, and the stock fan was busted.

[tom66]

Re: Validate my build

I will stick with the stock cooler. I've built a few PCs and always used the stock cooler. I've never overclocked them, but I have used them to play games and never had any issues. If I do encounter issues I'll order a cooler.

[mockingbird]

Re: Validate my build

Ok, just keep in mind that these pancake coolers are very much fan-dependent. It's not a big deal. If for some reason there's a problem with the fan, you'll know it, because your chip will throttle the speed down significantly.

[tom66]

Re: Validate my build

That's the same for most coolers. Very few heatsinks can dissipate heat effectively without forced air flow. I had a passive CPU heatsink in my old media PC and it was a massive, 90% copper beast. (It still had cooling via the PSU fan though that was on a thermostat which meant most of the time it did not turn at all, silent PC.)

[mockingbird]

Re: Validate my build

There's forced air flow and then there's forced air flow. These heatsinks are so thin, they're practically rudimentary in nature. It's more like a heatspreader than a heatsink. Absolute rubbish, and I would not use them on something that gets that hot.

[tom66]

Re: Validate my build

I really wish heatsink manufacturers specified °C/W figures instead of vaguely relying on design or size to sell them. It would make it much easier to shop for heatsinks. Luckily some review websites have started testing for this figure, but it's not exhaustive.

I think AA batteries, etc. should also have mAh/mWh figures on them (for non rechargeables) but then customers might realise that all AA batteries are virtually the same.

[tom66]

Re: Validate my build

PC built and all works good. In the end I went for an EVGA power supply because I saw a teardown showing all-Chemicon output capacitors. But I peered through the vent and I saw Capxon! Blugrh!!

I'm debating whether or not to send it back. It's got a 5yr warranty which I suppose I could use if the capacitors failed. And it was quite cheap for an 80 PLUS Gold PSU (650W). It's also modular - hard to find one like that around the same price bracket. Decisions, decisions...

[mockingbird]

Re: Validate my build

Which model?

I'd send it back.

[ratdude747]

Re: Validate my build

Sounds like evga pulled an Antec!

[tom66]

Re: Validate my build

To it's credit, there are some Chemicons on the output. But I can count at least four Capxon which could be output caps. It's a NEX650G1. Internal photo that I saw: http://www.overclock.net/content/type/61/id/1948589/

I don't mind the group regulation and I'm not using more than about 250W in normal use.

[mockingbird]

Re: Validate my build

I'd leave it alone, since it's only a 250w draw.

At leased they used proper RTV silicone and not organic glue.