Search results

gotta give the guy credit for successfully trolling youtube commenters.

the batch file for linpack doesn't take that long to execute. if you ran linpack manually setting it to loop a problem size of say, 25000 or 30000, for hours it would probably fail too.

i've found the quickest way to get the heat up is to not use the batch file but instead run the .exe directly in interactive mode. double click .exe, skip the first prompt (press enter), problem size -> 25000, leading dimension -> 25000, number of trials -> 1, data alignment -> 4 one trial...

same here, always an error after every test (if you uncheck 'stop on error'). tested on multiple systems, some that aren't even overclocked. probably a bug between linx and the latest version of linpack.

haswell does this by design, even if you're running at stock clocks and stock voltage. the only way to prevent the extra voltage juicing is to set a fixed manual voltage.

is your motherboard running all cores at maximum turbo? remember, intel specifies that this cpu should be running at 3.7 with all cores loaded, but ultimately this is up to the motherboard manufacturers to implement. a lot of them 'cheat' and run all cores at 3.9 instead. this in conjunction...

it looks like edge to edge glass, but it's really not. none of these slim 'bezelless' ips panels use actual glass in front of the panel, it's thin plastic (probably polycarbonate) and because of the way these panels are constructed they superficially look like edge to edge glass in photos...

it depends on how the voltage is set. if you set voltage using a fixed manual voltage, the cpu will always be running at that voltage regardless of load or p-state. similarly, using a voltage offset will increase voltage by the specified amount everywhere. adaptive voltage is the only one...

i was rolling in the aisles when i read that. this is reddit-worthy stuff.

this is possible, since when using adaptive voltage, the cpu will even more voltage when stress testing with a program that uses avx instructions. that means that if you run regular apps that don't use avx, the cpu is using the adaptive voltage you specified, which may not be stable at your...

aida64's fpu stress test is no joke when run by itself (you'll need to untick the other stress test boxes) - on my i5-4670k, it easily drove up temps 10°c higher over prime95 and linx. with the other stress test boxes ticked in addition to fpu, the temps don't get quite as high. on one run...

at idle, cpu drops into c1e (enhanced halt state), so temps and power consumption aren't affected much by voltage.

keep in mind though that on current intel cpu's, the socket clamp pushes down on the ihs/lid to force the cpu down against the cpu socket contact pins. if you remove the lid, you'll need to put some sort of shim in its place.

offset isn't going to help. setting a positive offset will just increase voltage everywhere. the only way to prevent that .090v jump when stress testing is to use manual voltage.

it's possible you may have hit a voltage wall at the multiplier you're trying to run. try dropping the multiplier a step and see if that gets you stable at 1.2 volts or lower. on my cpu, i had to push 1.255 volts (adaptive) to get it stable at 4.4, but it will run all day at 4.3 with only...

sounds like it will end up up being that way due to the 0.090v jump which you can't really control. is your cpu an i7? 1.175 is too low, thus the blue screens. try 1.180 and 1.185. it'll mean you'll end up running at 1.270v or higher when stressing, but there's not much you can do about...

yes, just watch the voltage measurement as the stress test runs, the highest voltage you see is the peak voltage.

why don't you start an occupy intel movement? as history as shown, occupy movements are very effective in instituting widespread change. never underestimate what six people united as one can do.

especially the more recent versions of prime95 that use avx. older versions that don't use avx won't stress the cpu as hard. linx with the latest linpack binaries seems to be even more demanding than prime95. my system that was running prime95 stable for hours blue screened within five...

becuase the 4765t only runs at 2ghz, turboing to 3. so of course the tdp is lower, its nominal as well as maximum frequency is lower than the other i7's. but for the same load and clock speed, there probably won't be much if any difference in power consumption between the 4765 and 4770k.

voltage offset increases or decreases the voltage by a fixed amount over the cpu's entire speed range. the other number sets the target voltage only for when the cpu is operating past its highest specified turbo frequency (i.e. overclocked). for example, my i5-4670k has a max turbo of 3.8ghz...

how did you test for stability? if that 1.255v is linx/avx or prime95/avx stable, then start with an adaptive voltage of 1.205. stress test some more and adjust the voltage depending on stability.

you mean your cpu isn't clocking down when idle? make sure speedstep is enabled in the bios and in windows set the power plan to 'balanced'. 'high performance' will lock your cpu in turbo mode. also, setting a manual voltage won't allow the cpu to reduce voltage when idle.

cache has its own separate voltage setting, you'll need to up that when you overclock the cache. of course doing that increases heat output.

thanks for the info. have you tried using the monitoring utility that came with the motherboard? or is it an intel motherboard? the aida64 cpuid tool seems to be fairly accurate. at least, more accurate than coretemp and cpu-z.

i didn't watch the video, but the reason why they don't advise using adaptive voltage with prime95 might be because the cpu can actually pull significantly more voltage than the specified adaptive voltage in certain situations (e.g. benchmarking programs that use avx instructions) when...

which motherboard do you have? i have the same cpu on an asus z87-plus motherboard, and the default 'auto' voltage setting is actually adaptive/1.250v. it doesn't tell you in the bios, but it does in the ai suite iii software. and under prime95 stress testing, the actual voltage ends up at...

i meant modifying the heatsink mounting bracket, not the motherboard. but i forgot the backplate would have to be modified in addition to the bracket and one will be more difficult than the other, so never mind.

are you guys overclocking the cache as well? (l3 cache and ring bus are decoupled from the core on haswell, so upping the core clock will not automatically overclock the l3 cache).

depending on how much material there is on the bracket, you might even be able to 'expand' out the existing lga775 holes to accomodate the 115x spacing. you'd only have to do about 1-2mm per hole. this is actually what cooler master does on their universal 775/115x/1366 brackets.

it just means that the images were taken using srgb color space and you have color management configured correctly. if color management were not configured correctly, the hp would have oversaturated and inaccurate colors. if the images were shot in adobe rgb (and assuming whatever you were...

oddly, only the 27" version is glossy.

the vg27ah is passive 3d, not active. so no 120hz and no lightboost.

it's possible. vga is an analog signal, so ultimately the whitest white the monitor receives (thus contrast) can be affected by things like cable and video card dac quality which aren't issues with digital connections. test it with a digital connection. eizo would probably ask you to do the same.

you're right about the philips having passive 3d and possible visible lines for some people. it's being sold on amazon, but from a third party vendor called 'fusion imports' so it's probably safe to assume it's not 'officially' available in the u.s. yet. the 'inner bezel' is actually part of...

we call this 'semi glossy'. details of objects that are closer to the screen can be discerned, like a hazy mirror, but objects (and light sources) that are further back end up looking smeared like on a monitor with a conventional matte coating. technically, the coatings on eyeglasses are...

if it looks fine now, it should stay fine.

that's rtc (overdrive) overshoot. just be thankful that your monitor lets you adjust the overdrive level, a lot of monitors don't.

the acer will be no different. the inner bezel on all of these 'frameless' ah-ips panels are the same deep black color. my viewsonic vx2770smh that i run with a .12 cd/m2 black level, still looks grey compared to the inner bezel. back when i reviewed the acer, it was the only glossy 27" 1080p...

on the acer, the overdrive setting reverts back to 'on' every time it loses sync (power cycle, resolution change, waking from sleep, etc.). and since the overdrive setting is in the service menu, and to enable the service menu you have to hold down the 'acer' button while powering the monitor...