• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

New HD4870 - runs very slow?

Fafnr

Junior Member
Hey all,

I just bought a Radeon 4870 since my 3850 was beginning to feel old and slow...
However, to be blunt, it's running like crap.

I've tested Far Cry 2 on 1024 x 768 (high on all settings) on both Windows XP 64 bit and Windows 7 beta, and it runs like a bloody slideshow. It's horrible.
I feel I should be able to run Far Cry 2 with this card?

My setup is as follows:
Intel Core 2 Duo E6400 (2.13 ghz)
4gb DDR2-1066 RAM
on an Asus P5K SE mobo, bios 1104 (According to CPU-Z)
Sapphire Radeon 4870 Toxic Edition
2 x 19" monitors attached to the 4870
couple of harddrives, etc.

Could this in any way be related to powerissues?
My PSU might be underpowered, but I've experienced no artifacts or crashes at all.

Both the XP 64 and Windows 7 were already installed. That is, no new installation just for the new gfx-card was done. (I also upgraded the RAM while I was at it)
Could this be the issue?

I've also noticed my CPU sometimes runs hot (about 75 degrees celcius) but the vendor where I bought my PC says it's ok.

Really, I'm looking for anything and everything that I could possibly do to help me fix this!

Regards,

Søren Andersen,
Denmark

(Edited for details on PC)
 
Use CPU-Z or GPU-Z to check your PCIe link width. If it's less than 16x, there's a problem.

Also, 75C is way too hot for that CPU.
 
The Link width is indeed 16x.
Is the CPU temperature related to how slow it seems, or just a problem in the long run? I'd kindda like to focus on 1 issue at a time. 🙂
I believe the cooler is an Arctic Alpine 7 BTW - if that matters at all...
 
No, I havn't enabled Overdrive for manual control over the fan.
Should I?

Since you asked, I've toyed with it a bit, and I can certainly hear it when I set it to 100%, but not really otherwise.
 
If your CPU is throttling, that could be causing performance problems, especially in demanding games. Check that your cooler is seated properly.
 
that might keep the heat down and the card performing at better temps - it could possibly be the problem...

I doubt it - I've yet to see it go above 50 degrees Celcius.
Not a bad guess though - but I feel like my card just isn't running at full speed to begin with...

If your CPU is throttling, that could be causing performance problems, especially in demanding games. Check that your cooler is seated properly.

Hmm... That seems more likely. Is there any way I can test this? Other than buying a new cooler, that is... I'd prefer to confirm it if at all possible?
EDIT:
I tried Far Cry 2 again with CPU-Z open on my alt. monitior.
I can see the following stats in CPU-Z:
Clocks (Core #0):
Core Speed: 1600 Mhz
Multiplier: x 6.0
Bus Speed: 266,7 Mhz
Rated FSB: 1066,7 Mhz

Is the core speed never rising above 1600 a sign of down-throttling? Or just something else entirely I do not understand? 🙂

Also, I've mentioned this before, I know, but could this is any way be related to lack og watt from my PSU?
I know it's somewhat underpowered, but I'd have expected it to act up some other way if that was the problem?

Regards,

Søren Andersen
 
Looks like you'll have to go into your bios and shut off the power saving features of your mobo. Or at least that's what it looks like to me by your multiplier.
 
Hmm... That seems more likely. Is there any way I can test this?

Reseat your cooler and make sure it's tightly mounted. If that doesn't help and you still have the stock HSF, try that one.

What are you using to monitor your temps? What does it say in BIOS? In Windows I suggest CoreTemp or RealTemp. Honestly, I wouldn't even leave my system on if it exceeded 60C for an extended period.


EDIT:
I tried Far Cry 2 again with CPU-Z open on my alt. monitior.
I can see the following stats in CPU-Z:
Clocks (Core #0):
Core Speed: 1600 Mhz
Multiplier: x 6.0
Bus Speed: 266,7 Mhz
Rated FSB: 1066,7 Mhz

Is the core speed never rising above 1600 a sign of down-throttling? Or just something else entirely I do not understand?

As GrumpyMan suggested, that sounds like SpeedStep. Look for BIOS options about SpeedStep and C1E and disable them. Do not disable CPU TM as this controls thermal throttling which your CPU should be doing if you're exceeding unsafe temps.

An insufficient power supply can cause instability and performance problems, surely, but shouldn't necessarily cause your CPU temps to skyrocket.
 
Ok, so - I've been following advice, researching a bit and stuff, and I think I've made some headway.
There's a combination of problems, all related to CPU temp.
First I tried running prime95 without ever chaning up from 1600mhz (speedstep).
Then I checked my cooler - it was definetaly loose. My CPU dropped from 75 degrees to about 50-55 idle when I reset it.

Now, when I run Prime95 I actually speed up to the 2.13ghz - nice! However, this only lasts for about 30-40 seconds, then the CPU temperature rises too high again (75+ degrees), and speedstep kicks in and drops it back down. (Actually, it oscillates a bit, and finally just stays at 1600).

To my mind, this means that my cooler isn't good enough. (I've checked that it's correctly connected to the mobo and all).
It seems very strange to me - it's a Arctic Alpine 7 - that it cannot cool a E6400!
However, it's the only explanation I can come up with.

So, I guess my last question is - what CPU cooler would you guys recommend?
It should preferably not be TOO big - I'd like to be able to move my case without risking too much mayhem...

Regards,

Søren
 
Originally posted by: Fafnr
Ok, so - I've been following advice, researching a bit and stuff, and I think I've made some headway.
There's a combination of problems, all related to CPU temp.
First I tried running prime95 without ever chaning up from 1600mhz (speedstep).
Then I checked my cooler - it was definetaly loose. My CPU dropped from 75 degrees to about 50-55 idle when I reset it.

Now, when I run Prime95 I actually speed up to the 2.13ghz - nice! However, this only lasts for about 30-40 seconds, then the CPU temperature rises too high again (75+ degrees), and speedstep kicks in and drops it back down. (Actually, it oscillates a bit, and finally just stays at 1600).

To my mind, this means that my cooler isn't good enough. (I've checked that it's correctly connected to the mobo and all).
It seems very strange to me - it's a Arctic Alpine 7 - that it cannot cool a E6400!
However, it's the only explanation I can come up with.

So, I guess my last question is - what CPU cooler would you guys recommend?
It should preferably not be TOO big - I'd like to be able to move my case without risking too much mayhem...

Regards,

Søren

Did you clean the old TIM off the HSF (using alcohol) and apply new TIM, before re-seating the HSF?

 
Did you clean the old TIM off the HSF (using alcohol) and apply new TIM, before re-seating the HSF?
No, I did not. I don't have any TIM here at home, so I skipped that...
I didn't think it was that important, to be honest?

But, if nothing else, it's worth a try I guess?

/Søren
 
For the sake of under $10, I always replace the TIM and clean the old stuff off when removing the HSF (or re-seating it).

The new TIM you use doesn't really matter (it will only affect the temps by a couple of degrees max. between brands), but the general favourite seems to be Artic Silver 5, however there are plenty of others out there. I use Akasa AK-455, because that is all I could source locally. It seems to work fine though.
 
Do you think it'll help enough to reduce the temperature by the 10-15 degrees celcius needed? I'm not that close to any place selling computer parts, so I'd really like to avoid several trips if possible.
 
Originally posted by: Fafnr
Do you think it'll help enough to reduce the temperature by the 10-15 degrees celcius needed? I'm not that close to any place selling computer parts, so I'd really like to avoid several trips if possible.

I'm not sure to be honest. Have you checked that the fan on the HSF is spinning?

The HSF seems to be man enough for the job according to Anandtech:
http://www.anandtech.com/casec...howdoc.aspx?i=3210&p=6

They tested it on a X6800 at 2.93GHz with a TDP of 75W, which is higher than the E6400 (2.13GHz and TDP = 65W). The Alpine 7 was 5C behind the TRUE with the X6800 under load.

Also, are you getting good airflow through the case (is your cabling neat and tidy)? Which case are you using btw?
 
Have you checked that the fan on the HSF is spinning?
It is. I've inspected it visually, and have a program checking it. It spins about 2k RPM, as it should...

I'm stumped too, tbh. I have no idea why it doesn't cool like it should.
I've checked all that I can - the cooler is on nice and tight, the power / temperature cable is on and working... It just generates soooo much heat? Is it possible there's just something wrong with my CPU?

The airflow should be OK.
I have a - I believe - 120mm fan blowing out air from right behind the CPU, and though there is some cabling, it's pretty far away from the case.
I've also tested with the case open, pulled out from under the desk where it usually is. It did help a bit, but not for long, and my CPU quickly overheated again.

Edit:
I took a couple of shots with my phone cam of the area around the CPU.
Can be seen here

I really have no good explanation for what's happening.

Oh, and you asked about my Cabinet?
It's a Thermaltake Mambo.
 
SpeedStep isn't what throttles your CPU when it gets too hot. SpeedStep is just a function that will reduce your multiplier and maybe voltage when your system is idle, then ramps back up on load.

Did you try using the stock heatsink? Also, again, what are you using to monitor your temps?
 
Sorry mate, I've ran out of ideas except...
1. Renew the TIM between the CPU and HSF (as I mentioned previously). However I can't see how this would decrease temps by 10-15C
2. Clean between the fins on the HSF (seems to be quite a lot of dust and crud on the HSF looking at the photos).
 
Originally posted by: CoinOperatedBoy
SpeedStep isn't what throttles your CPU when it gets too hot. SpeedStep is just a function that will reduce your multiplier and maybe voltage when your system is idle, then ramps back up on load.

Did you try using the stock heatsink? Also, again, what are you using to monitor your temps?

I don't think the HSF he is using is the problem. If you look at the Anandtech article I linked, the Alpine 7 performed significantly better than the stock Intel HSF and that was testing on a chip with a higher thermal envelope to the E6400.

The OP's case also seems to have reasonably decent cooling and cross air flow:
http://media.photobucket.com/i...ltakeArmor_TestFit.jpg
One 120mm dia fan in the rear (exhaust), another in the front (intake) and another 90mm dia fan in the top (exhaust). It also has a common internal layout.

It seems that the OP's CPU is thermal throttling (the fact that the multiplier dropped down as described in an earlier post), which means it must be hitting the 100C ish mark for this to kick in.

Either that or the CPU has dodgy temp sensors and it is thermally throttling prematurely. The only way to find this out would be to run a temperature / monitoring program (like CPU-Z / CoreTemp with Tjunction set to 100C) whilst the CPU is under full load (running Prime for example). Then you can see what the temps are and if its thermally throttling (CPU speed decreases).
 
I suggest:
Check CPU voltage - too high means excess heat.
Get fresh thermal grease (TIM) from a shop or online - it's worth the effort to help rule it out.

 
I'll try to answer everyone in this one message. 🙂

Also, again, what are you using to monitor your temps?
I'm using Hardware Sensors Monitor 4.4 - the only program I could find that would run on Win 7. Asus own won't install, unfortunately. It does seem like a decent program though.
I've also tried CoreTemp now which is actually even better. It says my CPU goes all the way to 85 degrees under load. Insane. It also says "Tj. Max = 85degrees", whatever that is.
I do notice, however, som strangeness in the coretemp readings. Once it hits 85 degrees the core temperatures sudden change back and forth between 85 degrees and -42! This could just be win 7 messing with the program, though


Did you try using the stock heatsink?
Can't - don't have one...

Renew the TIM between the CPU and HSF (as I mentioned previously). However I can't see how this would decrease temps by 10-15C
Yeah, that'll be my first try today together with cleaning the cooler.
Compressed air should do the job for cleaning the cooler I suspect?

Check CPU voltage - too high means excess heat.
It's 1.2250v @ 2.13ghz, but I can see it drop to 1.16v @ 1600mhz
I looked it up on wikipedia, and that seems about right?

 
You should probably RMA the CPU. I had a similar problem with an old P4D, defective CPUs are rare, but they can happen.
 
So, I bought some new thermal paste, cleaned off the old stuff, put the sink back on and... Everything works! :-D
Well, not quite that simple - on the first go, I either didn't have enough paste on there or I had seated the sink poorly or something, because it was a trainwreck...
Second time I added a bit more paste and turned the sink, and now it's brilliant!
My CPU cores idle around 53-54 degrees and max out at about 63. Still "too hot" but at least not dangerously hot. Gaming etc. is - not surprisingly - at lot more fun when your CPU isn't constantly shutting itself off in panick... 🙂

Anyway - thank you EVERYONE for your suggestions, help and time!
I'd never have gotten stuff working on my own without you guys!

Regards,

Søren
 
Back
Top