|
On April 19 2014 04:29 Ropid wrote:
Try disabling hardware acceleration. It will then draw everything on the CPU instead of using the graphics card and everything should look fine.
You get to the screen with that setting by opening Options and selecting Advanced at the top right.
Seems to be working so far, thanks a bunch. If it starts bugging out again I'll look into skyr's solution.
Thanks
|
On April 19 2014 16:32 Incognoto wrote:Show nested quote +On April 19 2014 16:27 wptlzkwjd wrote:On April 19 2014 15:41 HandA711 wrote:
Should I downgrade from Win 8.1 to XP with this rig? I play SC2, Dota 2, CSGO and emulate some PS2 games. I don't have any intention to play newer games with this PC to be honest.
Pentium E6500
Nvidia GT240 GDDR5 512MB
2GB RAM DDR2 Not sure why you would want to downgrade. Do you think using 8.1 will slow your computer down while running those games? o.O Well XP is friggin ancient
And amazing <3. I miss xp.
|
I recently did a clean install of Windows 7 (64-bit), and now I can't minimize Brood War using alt + tab. I play in windowed mode, and the game remains in front of the screen, blocking everything unless I actually click the minimize button. Whichever application I tab to becomes active; I just can't see it. Any idea what's causing this?
|
Hi there, is 68 degrees celsius dangerous for an i5 4570? Core#0 was 68 and Package was 68. The other cores were lower 60s.
Edit: sorry should've added that it's hot where I live and only getting hotter. It was a high of 88 and my a/c is not yet installed.
|
Around 70c is normal for load.
|
Sometimes one core will get hotter than the others because it's working and the others aren't. High 88°C is a bit high imo, 68°C is fine.
|
Pretty sure 88 was referring to a high of 88 fahrenheit in his home because he doesn't have air conditioning installed yet.
|
ah my bad should have realized that.
google told me 88°F is 31°C.
lol must suck living somewhere so warm
|
United Kingdom20345 Posts
On April 24 2014 16:32 Incognoto wrote:
ah my bad should have realized that.
google told me 88°F is 31°C.
lol must suck living somewhere so warm
yea with little airflow in a case with a GPU, those kinds of temps often mean 35c or even like 40c case temps which is pretty massive temp change vs 20
Sometimes one core will get hotter than the others because it's working and the others aren't.
As well as this, there is often a pattern with core temperatures on Haswell CPU's - on mine, for example, it can be like:
67, 68, 69, 62
and in another test - 74, 75, 76, 69
the last core being cooler is really common, and it's completely expected for them to be varied unless you're loading them all to 100%
http://www.overclock.net/t/1411077/haswell-overclocking-guide-with-statistics/10980_30#post_21941059
^This is good test for checking temps, if you grabbed the 295MB download and ran it with 8 threads (8 threads for i5 quads, 16 for i7 quads) but might get warm for you. If it's too hot by a significant margin(as it basically just shows you temps from solid video encoding, as hot as it could get) then you might want to remount the stock cooler (but i don't think much can go wrong there) or buy a basic aftermarket cooler which would be quieter for you than a stock cooler which is always spinning fast because high temperatures, and significantly better at cooling
it's ok to let it run at ~80-90c, but if it's sat at 100 (any task, really) then stop it
|
I downloaded the program and decided to give it a try. I'm using the stock cooler on an i5 4670 (no overclock), this is what I'm getting with the first loop 15% done:
![[image loading]](http://i.imgur.com/DYaIOmd.png)
Interestingly, I'm getting one core a couple degrees hotter than the others, rather than one cooler than the others. Also somewhat interesting is that I can't hear my stock cooler. I'm not joking, there's no audible difference to me between the stock cooler working at 70°C and the stock cooler at idle (which is roughly 30-35°C). I do hear my video card running at higher speeds though and the PSU is also somewhat noisy. I did delid my CPU though, a long time ago.
E: Final result, not too sure what to make of it except that temperatures seemed to be quite acceptable:
http://imgur.com/ixoa9Y0
|
United Kingdom20345 Posts
On April 24 2014 20:56 Incognoto wrote:I downloaded the program and decided to give it a try. I'm using the stock cooler on an i5 4670 (no overclock), this is what I'm getting with the first loop 15% done: + Show Spoiler +Interestingly, I'm getting one core a couple degrees hotter than the others, rather than one cooler than the others. Also somewhat interesting is that I can't hear my stock cooler. I'm not joking, there's no audible difference to me between the stock cooler working at 70°C and the stock cooler at idle (which is roughly 30-35°C). I do hear my video card running at higher speeds though and the PSU is also somewhat noisy. I did delid my CPU though, a long time ago. E: Final result, not too sure what to make of it except that temperatures seemed to be quite acceptable: + Show Spoiler +
Nice data! What's room temp like? You're probably somewhat limited by the very low heatsink area of the stock cooler, so delid wouldn't help as much as it would with a better cooling setup (like a midrange air cooler or better) where the main issue is getting the heat into the heatsink, instead of the typical "cool the heatsink faster"
also of note is that my CPU is encoding 2.3x faster than yours lol
Here's my temps with 1.265vcore and HT on, which is about 10c hotter than HT off
![[image loading]](http://i.imgur.com/V7luUNt.png)
was getting 4.04fps low priority
The initial high temps is because room was hot, i opened window at start of test, and also the temps that i reported before were a bit wrong because my room was colder than i thought. I got a drop from new case especially on extended testing, but not really a 7c drop.
Actually.. 2.3x is a way too drastic difference. I'm not sure what could cause it to be so bad, unless you were actively using the CPU for other stuff. It's not really primarily a benchmark anyway, just CPU temperature/stability testing
|
x2.3 is actually really quite a lot. You're overclocked to 4.5 GHz, with hyperthreading, am I right? I'm suddenly kicking myself for not looking, I'm not sure if the CPU was at 3.4 GHz or 3.6 GHz during the run.. I'm guessing 3.6 GHz though. So you have at least a full GHz advantage on my CPU + hyperthreading. It makes a substantial difference, as it should I suppose.
Rooms temps I didn't check but they were probably about 18°C. The delid remarks makes sense to me, delid helps transfer heat from the CPU to the heatsink; but in my case the heatsink isn't getting rid of heat fast enough for the delid to be truly efficient.
Like, a processor without a delid might produce 70W of heat and run at a higher temperature.
The same processor with the delid might produce 90W of heat (this numbers are arbitrary), so it would in theory run at a lower temperature (assuming same workload / voltage and same heatsink).
So if my low-end stock cooler can only exhaust 60W of heat, then the delid didn't really do anything to improve temperatures.
If the heatsink is high end and is capable of exhausting 120W of heat, then the delid did in fact help exhaust 20W more heat, which is what people are looking for when they delid.
Makes sense to me.
E: damn, nice performance on the 4770k.
Hmm x2.3 is a lot you say? I'll try to re-do the benchmark with almost nothing else running, I'll see how it goes.
Actually what are good CPU benchmarks to look at? forgot that this was a stability test.
|
United Kingdom20345 Posts
x2.3 is actually really quite a lot. You're overclocked to 4.5 GHz, with hyperthreading, am I right? I'm suddenly kicking myself for not looking, I'm not sure if the CPU was at 3.4 GHz or 3.6 GHz during the run.. I'm guessing 3.6 GHz though. So you have at least a full GHz advantage on my CPU + hyperthreading. It makes a substantial difference, as it should I suppose.
It shouldn't be that big of a gap
With a certain CPU and cooler for example, you might say that it can balance an 80w power load with CPU being 40c hotter than the air surrounding the cooler/radiator, or a 100w power load with CPU being 50c hotter than the air, something like that, not sure on scaling (if it's linear or exponential etc) but that should be pretty basic thermodynamics. In short, delid helps to utilize higher heatsink area etc, but does not really do much if the heatsink is already warm-hot and is the limiting factor
edit: CPU benching.. This should be good, it's just not really tuned to reflect real world encoding performance, but to max out cpu consistently. You could use something like Cinebench r15 - my scores there @4.6ghz were 952 multi and about 185 single with HT on, but HT scales very aggressively for the multi section there*, it's a bit RAM speed dependent etc. Best to run benches like Cinebench at high or even Realtime priority. The x264 5.0.1 benchmark that seems quite popular for those testing the encoder is actually trash, using a 2-3 year old encoder version with no Haswell optimizations or avx2 support etc
*Without HT, with 185 score on a single core you'd be expected to score about 4x that or a hair under for a true quad core. That's ~740 though - quite far away from 952. It's almost 1.3x scaling in performance from Hyperthreading, so as a benchmark, cine is quite i7 dominated. Version 11.5 was bad for comparing AMD vs Intel CPU's, but r15 is much better.
|
A 2.3 factor seems pretty unbelievable in that video encoding test. Maybe something is going on with AVX and AVX2? Might this be a freshly reinstalled Windows 7 without the relevant updates that enable AVX or the test is a different version?
|
Quick q about streaming if you don't mind:
I am currently using a 25/2 cable connection for 50 bucks a month, and can upgrade to a 35/3 one with the same usp for 10 bucks more. Is this worth it for streaming sc2 (will I see a difference)? I tried having bitrate as 1400 and noticed some slight delays in commands issued currently.
|
This is Windows 7, the build is 7600 and I've never installed updates on it. There's an update called KB2670838 which can mess with a game I play so don't install any updates at all, since things are running fine anyway. Maybe that's why.
Otherwise my Cinebench R15 score is 544.
E: [IMG]http://i.imgur.com/9DkImIh.jpg
By the way, Cyro you mentioned something about "high" or "realtime" priority. I couldn't find options like that in cinebench, so not sure what you're getting at. hmm
|
On April 24 2014 23:31 Incognoto wrote:
This is Windows 7, the build is 7600 and I've never installed updates on it. There's an update called KB2670838 which can mess with a game I play so don't install any updates at all, since things are running fine anyway. Maybe that's why.
Otherwise my Cinebench R15 score is 544.
Okay, so that's what's going on.  It's not that important for nearly all normal stuff, like you surely will not see any difference in games, but video encoding does actually get something out of AVX.
+ Show Spoiler +If you are interested why there needs to be an update to enable AVX, it's because the CPU core has a new set of registers on which those AVX instructions work on. The "kernel" of the operating system needs to know about those. The kernel's job is to stop program threads and load different ones onto the cores. Whenever it switches a thread somewhere, it needs to save all registers of that core into memory and then load the values that the other thread was using. So because the AVX instructions work on a new set of registers, enabling AVX for programs needs a change in the kernel of the operating system.
There might be an easy way to try out the newest updates for Windows without having to reinstall anything if it goes wrong. Perhaps if you manually save a restore point with the "System Restore" feature, it is able to revert any change made? I don't know if installing several years worth of Windows updates will destroy that restore point.
Another method is, you could make a system image backup of your C: partition (or the whole first drive if it's small like an SSD), save the backup file onto a HDD partition somewhere that's not C:. You could then restore that after things go wrong. The free versions of the various backup programs should be able to do it. I'd recommend putting the restore disc image of Acronis True Image onto a USB stick and then restart and boot from that USB stick. That boot disc has pretty much all features you need, can make backups and restore those, and you'll do it perfectly with Windows being shut down and offline.
Windows also has a backup tool that can make system images built in, but that one is super slow and the way I remember, you don't really know exactly what's going on, what partitions exactly are copied and what will be restored and overwritten.
|
I like the second method you described best if I were attempt to install possibly nefast updates, however I've been somewhat tramautized by the trouble I went through a few months back when I got my new SSD and tried to install windows on it, things like AHCI, disk partitions and BSODs drove me half nuts. I finally got around to getting it to work but not after at least a days worth of troubleshooting it. :p
I would be interested to try to install the windows update if only to bench and get a proper x264 bench but atm I don't feel like going that far out just for a single bench (which is a stress test anyway).
good stuff though
|
United Kingdom20345 Posts
On April 24 2014 23:30 coL.hendralisk wrote:
Quick q about streaming if you don't mind:
I am currently using a 25/2 cable connection for 50 bucks a month, and can upgrade to a 35/3 one with the same usp for 10 bucks more. Is this worth it for streaming sc2 (will I see a difference)? I tried having bitrate as 1400 and noticed some slight delays in commands issued currently.
If it's an improvement to your solid upload speed, probably
As for Incog: Ropid is 100% correct here. You should enable AVX and AVX2 and try again, you're losing a significant amount of performance without them. This x264 test >should< run basically like a benchmark - all you're doing is encoding the same file in the same way with two different CPU's or sets of settings, which is basically what any benchmark is - but i'l grab some HT off numbers to confirm that for you and compare, as well as check HT scaling.
You only need service pack 1 i believe, you don't need to download 50000 updates, and it's not just for a benchmark - you will impact the performance of numerous tasks on your system if you have important CPU instructions disabled. You can see here how much video encoding performance is slaughtered (which is basically the only CPU load while streaming, for example)
|
Hmm is that so..
well I'll look into that then. Not sure how it'll play out, I did some research and I think I found the appropriate service pack which doesn't seem to be the one I need to avoid. i'm going to DL and if it doesn't work out i'll just reinstall windows like i've done so countless times in the past, it's not something i'm afraid of doing, especially now that I have an HDD for backing up the files I don't want gone.
|
|
|
|
|
|
EPT
