General computer talk/advice

Its because his CPU isn't legit afaik, its a pre production or sample chip.

So it is a legit Intel CPU, it just isn't legit legit.

 

Damn, my single core is much lower than I expected, however I'm not too sure how accurate it really is.

Yeah its an engineering sample CPU, thats why it was $270 and not $1700

Only downside being its clocked lower at 2.2ghz instead of 2.4ghz.

 

To be honest you would expect the single thread to be between 14 and 28 times slower than the multi thread result depending on how much it throttles for multithreaded clock speeds.

 

That is problem with low clock rate, single core performance gets lower, in BeamNG that means earlier limit when running lot of lights or map with lot of objects.

With that good multicore score, many vehicles becomes possible, you can probably run lot more than 8 T75 on pure grid map with lowest details and 720p, so there are different sides in everything

If you want to know how single core test result translates to BeamNG, you can do this test, it is shadow calculation on objects that run on single thread or something other light and shadow related that loads single core and adding extra spots to off road roamer makes really lot of load to that single thread, that is why fps dies on this test, might explain if you have experienced odd slowdowns on some maps:


What I'm not exactly sure is how different GPU driver overhead and features affect this, for example if GPU can do X number of lighting calculations and this loads it to X+Y and thus different GPU might have effect on it too, but would need lot of different GPU's and CPU's tested on very exact same way to figure out more about it.

Long time ago there was feature called T&L and then I had GPU that could do 7 lights, which was lot back then, if there was more than num of lights it could do, fps fell through floor.

 

Yeah thats what I would have thought but even then my score is much lower than even 14%, clock speeds seem to stay pretty much steady at 2.2ghz on the single thread test. I think it must be off because my overclocked FX 8350 I used to have would be able to pull 95-100 in single threaded load and it was unable to run some of the heavier vehicles in BeamNG while my xeon has no problem with anything I've tried.

Yeah I can run at least 13 T series trucks, however the game wont load anymore without crashing so I cant test further.

I also tried that test out a while ago and didnt have any problems with it, so I'm not sure on that one either.

 

Perhaps its just better delegation of non game tasks to other threads?

 

I wonder, if you installed that chip into a motherboard that has overclocking capability if you could overclock it via the FSB. If it would even boot at all.

 

FX had probelm with certain type of calculation which reflected in to BeamNG performance, it is bit odd that Xeon shows so low result, but maybe it just don't like to math Cinebench offers.

If not having problem with running two of those roamers with added spot lights high beams on with same settings, then all I can think of is that for some reason GTX 1050 gets overwhelmed by number of light calculations and pushes those to be done by CPU while some other cards manage more light calculations and do it by GPU instead of CPU, all in all, that is very big mystery in BeamNG.

Going by Cinebench score, I should have 1.6 times more CPU power for that graphics thread to run, which means I should have higher frame rate than machine with faster GPU but slower single core result.

However I have not found information of what feature it could be that for example 970 has so much more than 1050.

What is interesting that you can run 13 T-series, while in Ryzen thread it was 11 T-series when fps was around 30, I guess multicore result is close to same, but do you get more or less fps with 11 T75 on small pure grid with lowest details? That information might help guys to have more accurate picture of how well Ryzen runs Beam, I would guess it being bit better than your CPU because slightly better multicore result.

That would be indication that Ryzen is as good in Cinebench as in BeamNG which I think is true and great achievement by AMD.


I would need to find 970 or 1070 to test so I could find out more about this high single thread load issue or someone with 1050 card to do exact same test for control check. Fun fact, with lights off or shadows off, I get barely 50% single core load on same spot and same settings. Can't really run glass top mountain with shadows at all because of this.


Some Xeons overclock a lot and I have seen fastest single core performances coming from some Xeons that has had crazy overclocks, not sure how those are done though.

 

All X99 boards have overclocking ability, the problem lies within the Xeon CPUs themselves. Most users can't get more than a couple hundred megahertz out of them max. I'd imagine it's the way they're binned/made.

However the LGA 1151 Xeons show some nice overclocking ability, as did the X-series LGA 1366 Xeons.

 

Saw a ThinkPad being advertised as a "Fast Gaming Machine!" on eBay.

Needless to say, it had a Core i5 and integrated graphics.
Core i5 m520. It's hardly a fast gaming machine...

Goddamn eBay.

 

Today I decided to try learning basics about overclocking and maybe try something.
I read an article on TomsHardware about temperatures and decided to do some testing on stock clocks. My CPU cooler is rated at 100W and and now CPU is about 83*C hot(that is quite a lot, I know)
I guess that means no overclocking for me.

Question: If voltage stays the same, but clocks increase, do temperatures also increase?

 

Nope not until you get a good cooler. Yes the heat generated will increase. Also, I don't know what power supply you have but make sure you have a nice good quality power supply with stable voltage rails. A cheap power supply can hold you back overclocking.

 

@Rolph
A hyper 212 evo is a decent cooler but there's a few decent ones, check the reviews floating around various orifices of the interwebs for good coolers. I have a 4790k that couldn't even hold 4ghz on it's stock cooler, so I plunked one of the best coolers I could find on it, and atleast it can hold it's turbo mode @ that. These processors aren't soldered to the heat-spreader (metal top of processor), so some are worse than others. Mine is awful. These cpus are near the top end of what they can be, but a decent cooler can help you substantially speed that up (well, more like 15% increase, possibly 20% in ghz speed). I wouldn't realistically think to get more than avg 4.2ghz out of it on a good cooler. I personally don't do water cooling, but it's possible.

I have some red Phanteks 140mm dual-tower cooler, lots of room and hard as heck to mount with man's hands. Do read reviews! Not 100% the best unit out now, Noctua D15 is good IF it fits in the case.

 

I have this PSU - Fortron FSP HYPER 600 PSU 12cm, Active, 8*SATA, 4*PCIe-6+2, 1*4+4pin(EPS), 2*4 pin Molex 85+ Black 600W
Wattage headroom is enough I think, but I know nothing about those advanced PSU things.

One MSI app offers Game Boost that changes clocks to 4.1 GHz. Might try it later while looking at temps.
Also during real life usage I haven't managed to get it over 70*C (at least haven't seen it)

I would love a better cooler, but my PC is prebuilt with warranty and changing cooler would most likely kill the warranty in all possible ways. If I could choose the cooler, I would never ever get this one, but overall the price for this prebuilt system was much better than other similar prebuilts in my country, so I had to take some compromises.

 

That power supply looks very dodgy and just looking at the internals of it I can guarantee it's not going to handle a 600w load. i'd say 400w at best. Oh and overclocking will void your warranty so yea... If you do want to overclock it though, I'd get a new PSU and like Bob said a 212evo. Best bang for the buck cooler out there.

 

Wouldnt bother overclocking at all, your temperatures are already too high. Way too high, you are already risking damage to system without overclocking
--- Post updated ---
Concluded that my favourite bit of new electronic gadgetry to be released in the last few years is definitely the nintendo switch.
Launch lineup weak sure and it doesnt even have much in way of non gaming functions. But whats there right now, works fucking great, its such a neat device.

 

FSP (Fortron source power) is a major OEM (like Seasonic is, the brand I use) and makes units for MANY psu manufacturers. Those are generally solid 'it's good enough' units, but that can always be looked up.
All the FSP-source unit's I have had have been compliant and decent.


Don't void the warranty of your PC. When it's up, get a bolt-down tower-style heatpipe cooler.

@Eastham If you have infos as to those FSP's being bad let me know. I always keep an open mind.

 

"CapXon" branded capacitors, known for going bad, and those heatsinks seem a little whimpey for this 700W rated model.


--- Post updated ---
I will concede though. They do have some pretty beefy well made looking units too. Nippon Chemicon and Nichcicon caps.

 

Why your fan reads as 0rpm?

I have been happy with Thermaltake Macho Rev.B, I have only one fan in it, but I just did Cinebench CPU test and temps topped out as 40C, of course there is no way to overclock i7-6700, but I will probably keep this cooler in next CPU setup too if possible.

 

Old bios for your board, or a modded bios, and bus-clock (BClock) that bad boy right along like they were doing with the i3's.

Yes, I've heard of/seen that happening (across the industry), usually when an OEM asks them for 1000+ PSU's with the cheap stuff, and they don't wanna spend a lot to get it to market. IIRC 'Ultra' (available on 'Tiger Direct') was one to do this often. I had to fix more than one of those! I thought they were past most of this mess from 10 years ago though - I could be wrong as I'm only running SeaSonic PSU's on my stuff, and EVGA junk if ya gotta go the cheap route... So I haven't dealt with Fortron as of late (unless, they make the EVGA units, hope not!).
FSP *DOES* have some good units (like you said) though, and even the half-decent ones will hold up pretty good if kept cool enough with a good power signal/on a good surge supressor. Atleast it's not one of those 8 ounce China units that blow up at half the rated watts.
Also, to add to this, contrary to popular belief, ANTEC, once renowned, has long been surpassed these days in quality. The cheap unit available at Staples in the USA is complete rubbish. 350 watt, wouldn't trust it powering a cat toy, might blow up my cat.
Edit: No sylvester for the moment, but daffy duck getting blown up in Speedy's Go Go Amigo will have to do.
Good pic for what happens when you use cheap power supply!