Computer issues of the various kind

[Terrik]

Alright, so some of you might remember that I had to get a card switched out (GTX 580M).

Well that's all done, expect for a small issue--I'm missing a 4GB RAM stick. So I went from 16Gb to 12GB--but thats ok, because It's going to be replaced.

That all being said: here's whats going on--in certain gamess (well, ok, fine, the only one I really play--DOTA 2) I've been getting weird slowdowns. It'll drop FPS/slowdown and then go back to normal. It'll do this every so often. My first immediate thought was the video card---but I'm not so sure thats it anymore.

Here's what we know:


I opened up Nvidia control panel and let it show the SLI indicator---every time the game slowed down, both bars went down--so it wasn't just one card that was underperforming.

Thinking it might have been new drivers (the newest NVIDIA having a new DOTA 2 profile), I rolled back the drivers. No luck.

I turned off SLI and got the same problem.

Here's another weird thing going on---my alienware laptop will flash the notification that I'm on AC power when I plug it in, but the icon shows it's not plugged in and draining battery--so I popped out the battery and tried to run on AC power alone, and that's what It's on right now.

Now that I really think about it, the slowdowns seem eerily similar to what happens when I unplug the AC power and run the laptop on battery.

Could by AC power cable be going bad and I'm not getting enough constant power to the system? Can these things go bad that way?

 

[Bubble181]

Well, obviously, this could easily still be 100.000 different things, and I'm not qualified to answer.
However, yes, power cable/batteries can and do go bust that way; had it happen myself as well. The battery was still good (well, not brand new but the laptop could run off of it for over an hour), the cable was good, but the power connector had a bad connection - the laptop would occasionally just decide to "lose" the power connection for a few seconds/minutes. Not a big problem while surfing, crashed quite a few games.
Try making a few adjustments to what your laptop does if it loses power (like, turn the brightness of the screen waaay down) and runs on battery, then play some more. If, during the slow-down, the screen also goes darker, you can be fairly sure it's a power supply issue, which may or may not be easily fixable (wiggling the power cord a bit, a new power cord,...).
If it doesn't, it's unfortunately probably another issue

 

[Ravenpoe]

Check your power saving settings and make sure it's not throttling your performance to save power.

Also, it's kind of a longshot, but ram likes working in pairs (or they used to... that could have changed in the last 10 years I haven't been keeping up with shit) so you could even try removing the extraneous 4GB stick (bringing you down to 8) until you can get the fourth replaced for a full 16.

 

[PatrThom]

It's also possible that the AC adapter socket (or the DC-IN port) are not making/keeping proper contact with the cord, or that the socket itself was bumped enough during the repair to loosen the solder connection to the motherboard.

On a separate note, having unequal-sized RAM in the different RAM slots can cause other issues if the computer expects to be able to interleave the two slots, which could account for slower memory access, but dunno how it would impact SLI (because it shouldn't unless the graphics cards themselves are set to use more than their onboard amount of RAM).

--Patrick

 

[Bubble181]

Also, it's kind of a longshot, but ram likes working in pairs (or they used to... that could have changed in the last 10 years I haven't been keeping up with shit) so you could even try removing the extraneous 4GB stick (bringing you down to 8) until you can get the fourth replaced for a full 16.

Not as true as before (I remember it was nearly literally impossible to run with 1 or 3 sticks), but yes, this might actually help.

 

[PatrThom]

nearly literally impossible to run with 1 or 3 sticks

This is usually in cases when the memory interface to the CPU is 256 bits wide, but the memory chips are only 128 bits wide each and so it requires both of them to make a full 256-bit channel.

--Patrick

 

[Terrik]

Here are some sceenshots that might help out or give clues. I'll find out if its the RAM when I get my new stick

Spoiler

SLI bar showing everything is fine and dandy. Smooth as silk.

And when it slows, both cards go down.

Then back up again.

Then back down again. So on and so forth.

System info: Power settings

Temps

NVIDIA control panel settings

 

[PatrThom]

It might seem silly, but you can test for RAM issues by taking out the extra stick (bringing you down to only 2). At least, if performance goes up, that'll prove it's an interleaving issue (assuming 8GB is still enough RAM to function for you), or if the performance goes even further down, that'll prove that it's because of the reduction in RAM.

--Patrick

 

[Terrik]

I got a new stick...it's a 1rx8 as opposed to the 2Rx8 I have but I don't think that will cause any issue (I told the guy the speed I needed DDR3 1600mhz...I assume thats what it is. It's not like its written on the card). Regardless, I started up Dota 2 and it still does the random slowdowns. I'm lost as to what to do now.

EDIT: I went into NVIDIA control panel and set DOTA 2's profile to SLI rendering mode: Single GPU to see if SLI was causing the problem---it isn't because the problem still remains.

 

[PatrThom]

It could be written on there. DDR3-1600 is the same thing as PC3-12800.
As a general rule, it's not recommended to mix different memory types, but if it works, it works.
Only thing that springs to mind at this point is to change the performance on battery from "balanced mode" to "max performance" to eliminate the possibility that it keeps cycling up and down due to the power thing.

--Patrick

 

[Terrik]

It could be written on there. DDR3-1600 is the same thing as PC3-12800.
As a general rule, it's not recommended to mix different memory types, but if it works, it works.
Only thing that springs to mind at this point is to change the performance on battery from "balanced mode" to "max performance" to eliminate the possibility that it keeps cycling up and down due to the power thing.

--Patrick

Oh, then yeah, I believe the speed is the same. Im pretty sure it was written as PC3-12800. It's a 4GB ram card like the rest of them, expect instead of having 8 chips, this one has 4.

I changed performance on battery from balanced to max under GPU power state, but the problem still occurs.

I also went into safe mode, ran driver sweeper to uninstall the video drivers and reinstalled the latest drivers and it's had no effect. I mean....what else is there left?

 

[Terrik]

So, a few parts replaced---there was an issue with the connector to the battery and somesuch....it wasn't really clear. Long story short, my computer is back. I think. I wanted to test it, so i ran 3Dmark vantage. Everything went fine. I got 21,000~ which is standard for a my system spects w/o overclocking. I checked my HW monitor to check temps and it said a CPU core or two hit 98*C. Knowing myself that that is completely insane, I kinda freaked out, but then again---can that even be possible? I'd assume the system would have slowed, or even shut off at those temps. The computer tech said he never touched my CPU. Could I be getting incorrect readings?

I also ask this because I'll see it hit, say, 90+ and then drop down to 60*C within a second or two which leads me to believe it is more likely it is NOT hitting that high because I don't know how you dissipate 30*C that quickly. I can see 70*C or so...but 90*C+??

 

[PatrThom]

That wouldn't be unusual if what you're seeing is your system hitting 100C for a split second until the fans kick in and bring it back down again. It is possible for the numbers to change that quickly, the size of a CPU die is really small, so small that it has very little thermal mass, and so it can ramp up and down in temperature fairly quickly.

Still, I would make sure that any air vents (or the paths to them) are not blocked by fuzz, or a cable, or whatever.

--Patrick

 

[Terrik]

That wouldn't be unusual if what you're seeing is your system hitting 100C for a split second until the fans kick in and bring it back down again. It is possible for the numbers to change that quickly, the size of a CPU die is really small, so small that it has very little thermal mass, and so it can ramp up and down in temperature fairly quickly.

Still, I would make sure that any air vents (or the paths to them) are not blocked by fuzz, or a cable, or whatever.

--Patrick

Would such short bursts of heat be damaging? When I told the repair guy it was hitting that high he told me it was impossible because my computer would have shut off. Then again, this is the same guy that took 3 attempts to fix my machine...so...

 

[PatrThom]

In short, yes. Heat like that will shorten the life of computer components. That's why you want to make sure the fans are working. The alternative would be to have your fans running on high all the time, which would always keep the temperatures down but would be very noisy. It's a balance.

--Patrick

 

[Terrik]

Almost 3 months in and my laptop isn't fixed. I have taken care of the heat issue by replacing all 3 heatsinks.

But now a weirder problem has arisen: My laptop AC adapter works perfectly fine----until I play games. Then it dies. It might work ffor just a little bit again if I unplug it and plug it in again once or twice. Again, this ONLY happens during games. If Im web surfing, or watching videos, it's fine. Charges the battery fine--but pop open Dota 2 or TF 2 for a few minutes and the little green light dies on the adapter. I mean...what?..I don't even....huh?

 

[PatrThom]

If the computer draws more power than the adapter can provide, it will probably shut down.
Power supplies do get weaker over time, so if it was operating close to the limit to begin with, it will eventually age to a point where it can no longer support max draw. It's really hard to tell this without some way to measure what sort of amperage is being drawn, though.

--Patrick

 

[Terrik]

If the computer draws more power than the adapter can provide, it will probably shut down.
Power supplies do get weaker over time, so if it was operating close to the limit to begin with, it will eventually age to a point where it can no longer support max draw. It's really hard to tell this without some way to measure what sort of amperage is being drawn, though.

--Patrick

So it IS possible. I guess I need to find an Alienware/Dell store around here and see if they'll let me test a placement AC adapter.

 

[Terrik]

So, while waiting for a new power cable to come in, I went to America and came back. Computer has not been touched in 3 days.

BEEP BEEP BEEP BEEP

BEEP BEEP BEEP BEEP

BEEP BEEP BEEP BEEP

I know that sound, RAM read/write failure. Right? Let's open it up...take out, clean, reset all 4 ram chips/slots.


BEEP BEEP BEEP BEEP


oook...

Hold power button without cable and battery in...




BEEP BEEP BEEP BEEP


ok ok...check the SLI cable..



BEEP BEEP BEEP BEEP.

Ok ok fine, where's my sledgehammer....

 

[PatrThom]

Time to start pulling the sticks out and putting them back one at a time to find the bad one, then.

--Patrick

 

[jwhouk]

I'm just having connection issues.

Since switching to a new router, my main computer has apparently started to have issues connecting via the LAC, and has been defaulting to the wireless connection when I've had the wireless router connected.

Anyone with any clue as to how to "clear out" network connections in Windows 7?

 

[Terrik]

Time to start pulling the sticks out and putting them back one at a time to find the bad one, then.

--Patrick

Four beeps turned to five...reseated all ram cards again...hold FN KEY...RESTART....running preboot tests (it's not beeping...holy crap)...running full memory test and then going to update bios, fingers crossed.

 

[Terrik]

Success? I reinstalled the bios...played a game of Dota 2...

Lets see...highest CPU temp Core #1 87C, Core 2 # 83C, Core #3 78C Core #0 81C

Highest GPU temps GPU #1 @ 81*c GPU#2 @ 74*C


That's.....that's not awful?


Nothing blew up (yet). I think those temps are livable. Nothing like the 90+C that I was getting before.

 

[PatrThom]

Those numbers aren't outside of expectations, at least, though of course it would be nice if they didn't stay there all the time.

--Patrick

 

[Terrik]

Those numbers aren't outside of expectations, at least, though of course it would be nice if they didn't stay there all the time.

--Patrick

They don't. When I was playing dota 2 they went up. They seem to drop,to 40-50c when not gaming

 

[PatrThom]

Normal running temperature for a CPU/GPU should probably be somewhere around 110-130F.

--Patrick