• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

"I need a New PC!" 2012 Thread. 22nm+28nm, Tri-Gate, and reading the OP. [Part 1]

Status
Not open for further replies.

Ocho

Member
My final build, gentlemen (changes from previous build in bold):

Case: Fractal Design Arc Midi
CPU: Intel i5 3570K 3.4
Motherboard: ASUS Maximus V Gene
Cooler: CM TPC 812
RAM: Corsair Vengeance Low Profile 1600
PSU: Corsair 650HX
GPU: EVGA GeForce GTX680 SuperClocked 2048MB GDDR5
Drive: Samsung F3 1tb (decided to purchase ssd at a later date)
Monitor: Asus VW246H 24-Inch Full-HD LCD Monitor

Any last opinions before I order?
 
Im ready for some damn NDA's to end!

7970 ghz edition (anytime now)

660/660ti/whatever (rumored the 25th? Not sure if that's still on)

Trinity desktop (Tom's did a review, but only compared to AMD and promised to compare to some Intel stuff, which they still havent posted)
 

cametall

Member

kharma45

Member
Looking to upgrade my 4890. I have a budget of about €300 any recommendations?

Cheers

7850 would probably be the best bet in that budget if you're happy enough to go AMD again, especially if you overclock it, and in saying that, if you are going to do that it's worth going for one of the ones with the good custom coolers like the MSI Twin Frozr III and avoiding the likes of the VTX with the stock cooler.
 
7850 would probably be the best bet in that budget if you're happy enough to go AMD again, especially if you overclock it, and in saying that, if you are going to do that it's worth going for one of the ones with the good custom coolers like the MSI Twin Frozr III and avoiding the likes of the VTX with the stock cooler.

Cheers, the MSI Twin Frozr looks perfect for me. Happy enough to overclock ^^.
 

kharma45

Member
Cheers, the MSI Twin Frozr looks perfect for me. Happy enough to overclock ^^.

That's the one I ended up with, had a Sapphire on order but changed to MSI because of the better RMA/warranty.

After unlocking the voltage this is what my card is sitting at

ihOQDpx9eoK4d.png

Most seem to overclock pretty well, it's more than likely you'll get a good one as most people have done in this thread http://forums.overclockers.co.uk/showthread.php?t=18389760
 

Effect

Member
Is it better to keep the graphics card fan set to auto or manually set it? Is there a risk of having it fail if set to manual?
 

MrBig

Member
Is it better to keep the graphics card fan set to auto or manually set it? Is there a risk of having it fail if set to manual?

If you want to do it manual you can fine tune the curve a lot better than they conservatively do at the factory. Don't just manually set a static speed though.
 

hypernima

Banned
I love my Gigabyte X79 a lot. Way more than I thought I would as a longtime ASUS fanboy. There is a lot of love for the ASRock as well though. What are you aiming to do with the build?

Intense video/3d work. I know ya suggested it but I saw this and considered it as well also.
Just don't know which is better.
 

RedSwirl

Junior Member
I think the last thing I haven't tried with this Wi-Fi situation (aside from re-installing Windows) is buying a range booster or high gain antenna or whatever.

Suggestions? I'd rather not buy a USB enhancer because If I did that, I might as well just buy a USB Wi-Fi adapter anyway, and I'm running out of USB ports as it is.
 

cametall

Member
TurboV from the AI software. http://www.asus.com/Motherboards/Intel_Socket_1155/P8Z77V_LX/#download

Utilities, then the one with all the programs listed in it, third from the top. You can also define your fan profiles from it.

set multiplier to 100mhz x 44. Test stability at same voltage. Then drop it as far as you can using TurboV

So I've been adjusting the vcore and I'm down to 1.26v with C States and Speed Step disabled (1.27 if enabled).

Anyways at 1.25v I get a BSOD code 0x0A which as I understand it means I need to increase the QPI/VTT. Is this worth doing or should I just be happy at 1.26v (4.4ghz)?

One thing I'm noticing in CPUZ, at idle Core Voltage = what I set the vcore to, but once Prime95 starts the core voltage drops by about 0.6. Is this normal or is something causing the vcore to drop further than what I set it to?

EDIT: I'm seeing this drop is caused by vdroop (you mentioned it earlier).
 

Hazaro

relies on auto-aim
My final build, gentlemen (changes from previous build in bold):

Case: Fractal Design Arc Midi
CPU: Intel i5 3570K 3.4
Motherboard: ASUS Maximus V Gene
Cooler: CM TPC 812
RAM: Corsair Vengeance Low Profile 1600
PSU: Corsair 650HX
GPU: EVGA GeForce GTX680 SuperClocked 2048MB GDDR5
Drive: Samsung F3 1tb (decided to purchase ssd at a later date)
Monitor: Asus VW246H 24-Inch Full-HD LCD Monitor

Any last opinions before I order?
Looks good.
How much would an unlockable 2GB 6950 go for? $200 or so?
160-200 depending on what/where you are listing and what other listings are up.
So I've been adjusting the vcore and I'm down to 1.26v with C States and Speed Step disabled (1.27 if enabled).

Anyways at 1.25v I get a BSOD code 0x0A which as I understand it means I need to increase the QPI/VTT. Is this worth doing or should I just be happy at 1.26v (4.4ghz)?

One thing I'm noticing in CPUZ, at idle Core Voltage = what I set the vcore to, but once Prime95 starts the core voltage drops by about 0.6. Is this normal or is something causing the vcore to drop further than what I set it to?
You mean 0.06V right :)
It's normal. The reason for this is to act as a buffer for fast voltage switching between states to help stability.

On newer boards there is a certain setting that can mitigate this, but it was originally put in for stability reasons to keep everything working as CPU changed load/speed/voltage.

Depending on how harsh the Vdroop is, you may or may not want to enable this feature.
If you are concerned about higher idle volts, you should be overclocking with offset if you are on SB/IB though.
 

cametall

Member
You mean 0.06V right :)
It's normal. The reason for this is to act as a buffer for fast voltage switching between states to help stability.

On newer boards there is a certain setting that can mitigate this, but it was originally put in for stability reasons to keep everything working as CPU changed load/speed/voltage.

Depending on how harsh the Vdroop is, you may or may not want to enable this feature.
If you are concerned about higher idle volts, you should be overclocking with offset if you are on SB/IB though.

Yes, sorry, 0.06V.

EDIT: So I'm running IBT to test stability and I'm up to 1.29v now. Still testing stability with it.
X2: IBT scares me. I did 10 runs on high and one core hit 82c. It passed the 10 runs though at 1.29v. Doing P95 overnight now to try that. I guess I just got a weak chip for OCing (not the 3570K overall, just the one I own)...
X3: Unstable at 1.29v too lol, crashed in less than 30 minutes of P95 starting.

When I bought the PC it came OC'd to 4Ghz and vcore 1.36v, which was very high for that clock speed. It looked like all that was changed was the clock speed and vcore voltage. Should I reset the BIOS to default and start again?
 

Hazaro

relies on auto-aim
Yes, sorry, 0.06V.

EDIT: So I'm running IBT to test stability and I'm up to 1.29v now. Still testing stability with it.
X2: IBT scares me. I did 10 runs on high and one core hit 82c. It passed the 10 runs though at 1.29v. Doing P95 overnight now to try that. I guess I just got a weak chip for OCing (not the 3570K overall, just the one I own)...
X3: Unstable at 1.29v too lol, crashed in less than 30 minutes of P95 starting.

When I bought the PC it came OC'd to 4Ghz and vcore 1.36v, which was very high for that clock speed. It looked like all that was changed was the clock speed and vcore voltage. Should I reset the BIOS to default and start again?
Don't run IBT dude. It's for crazy people.

Go 1.25V and 4.2Ghz and work up or down from there on speed and volts.
Which case do you guys think would be best for a gtx 690?
Pick anything over $60 and add fans from the OP.
 

sk3tch

Member
Don't run IBT dude. It's for crazy people.

Go 1.25V and 4.2Ghz and work up or down from there on speed and volts.

What's better for testing true stability? I mean, if you just want to say you have a 5 GHz overclock you can run Prime95 for 15 minutes and call it good...but IBT is the torture test to run. Even the standard test has passed for me - but failed with I run extreme. So I always validate my OC's that way. It may shorten my e-peen due to a lower OC but it removes any chances of blue screens.
 

abunai

Member
What's better for testing true stability? I mean, if you just want to say you have a 5 GHz overclock you can run Prime95 for 15 minutes and call it good...but IBT is the torture test to run. Even the standard test has passed for me - but failed with I run extreme. So I always validate my OC's that way. It may shorten my e-peen due to a lower OC but it removes any chances of blue screens.

AIDA64 is the best tool, imo. It's not free though, but there is a 30-day trial.
 

sk3tch

Member
AIDA64 is the best tool, imo. It's not free though, but there is a 30-day trial.

I will have to check it out. I've been using this for only a couple of weeks - but it's free and from Intel so I imagine it's results are super solid, as well. :)

http://www.intel.com/content/www/us/en/motherboards/desktop-motherboards/desktop-boards-software-extreme-tuning-utility.html

The real question is what duration is appropriate for OC validity? Some will be comfortable with a 10-15 minute run. But I argue it requires a bit more than that to show true stability. It doesn't have to be hours...but 60-90 min. would probably be good as a minimum baseline. I guess it comes down to personal preference and what you're trying to achieve with your OC - is it for a one time suicide benchmark run or is it for 24x7x365 OC?
 

abunai

Member
I will have to check it out. I've been using this for only a couple of weeks - but it's free and from Intel so I imagine it's results are super solid, as well. :)

http://www.intel.com/content/www/us/en/motherboards/desktop-motherboards/desktop-boards-software-extreme-tuning-utility.html

The real question is what duration is appropriate for OC validity? Some will be comfortable with a 10-15 minute run. But I argue it requires a bit more than that to show true stability. It doesn't have to be hours...but 60-90 min. would probably be good as a minimum baseline. I guess it comes down to personal preference and what you're trying to achieve with your OC - is it for a one time suicide benchmark run or is it for 24x7x365 OC?

For me? I just leave it on overnight. If I didn't do that though, 3 hours or so would be enough for a safe 24/7 overclock though. In reality, it's best to benchmark with what one actually uses the computer for. I've had chips that survive 2 day+ AIDA64 runs at 100% with no problem, only to bsod after ~10 minute in a cpu-intensive game. Real world testing is more useful, i'd say.
 

sk3tch

Member
For me? I just leave it on overnight. If I didn't do that though, 3 hours or so would be enough for a safe 24/7 overclock though. In reality, it's best to benchmark with what one actually uses the computer for. I've had chips that survive 2 day+ AIDA64 runs at 100% with no problem, only to bsod after ~10 minute in a cpu-intensive game. Real world testing is more useful, i'd say.

You've just disqualified AIDA64 as an OC validation tool, then. ;)

3-4 hours of IBT Extreme and if your CPU and OC survive - it's a lock.
 

Negaiido

Member
I've got 2 SSDs, one of 64 GB and the other one of 256 GB.
What should I do with them? Install Windows on the 256 GB and cache the 64 GB one or install Windows on the 64 one and install programs on the 256 GB one?
To get the best performance that is :)
 

Ocho

Member
In the end, I decided to spend a little less money and got the evga 670 GTX FTW instead of the 680 and grabbed the HX750 instead of the HX650. My logic is that I could get another 670 for SLI soon, or replace the card for a better one in the future, and a 750W PSU can last me 5 years if I don't go crazy quad sli.

Thanks everyone for their awesome help, specially mkenyon for his awesome assistance and Hazaro for the builds in the OP.
 
Hello smart and aesthetically pleasing PC overlords. I'm trying to help my nephew upgrade his PC and struggling a bit on the graphics card.

He needs to run three monitors... so that's the first concern. I have no idea if any of the cards I've been looking at will support that, or how to really find out. That aside, the ones I think fit the bill best are these:

7850
6970

GTX570
GTX560Ti (448)

These all seem to be quite similarly priced, and similar in power (from what I've seen the Nvidia cards may have the edge there). Any suggestions? Will any of those support three monitors? Does one stand out as much better than the others?

Thanks.
 

Smokey

Member
Hello smart and aesthetically pleasing PC overlords. I'm trying to help my nephew upgrade his PC and struggling a bit on the graphics card.

He needs to run three monitors... so that's the first concern. I have no idea if any of the cards I've been looking at will support that, or how to really find out. That aside, the ones I think fit the bill best are these:

7850
6970

GTX570
GTX560Ti (448)

These all seem to be quite similarly priced, and similar in power (from what I've seen the Nvidia cards may have the edge there). Any suggestions? Will any of those support three monitors? Does one stand out as much better than the others?

Thanks.

You need two of the same card to run 3 monitors with Nvidia's 500 series. You can run 3 monitors with one card on the AMD side. Not sure how the 7850 stacks up to the 6970 in gaming though.
 

scitek

Member
Welp, have to RMA my GTX 670. It just took my buddy 3 weeks to get a replacement from Newegg, is that the average turnaround?
 

kharma45

Member
You need two of the same card to run 3 monitors with Nvidia's 500 series. You can run 3 monitors with one card on the AMD side. Not sure how the 7850 stacks up to the 6970 in gaming though.

They're pretty close

http://www.anandtech.com/bench/Product/549?vs=509

But the 7850 runs cooler and uses less juice, plus it overclocks like an absolute demon too so would be my choice.

Something like the MSI Twin Frozr III 7850 would be a good buy, with a small voltage bump I've mine at 1200MHz.
 

mkenyon

Banned
What's better for testing true stability? I mean, if you just want to say you have a 5 GHz overclock you can run Prime95 for 15 minutes and call it good...but IBT is the torture test to run. Even the standard test has passed for me - but failed with I run extreme. So I always validate my OC's that way. It may shorten my e-peen due to a lower OC but it removes any chances of blue screens.
On the flip side, most OC blue screens after a P95 run that occur are during idle, not load. Ensuring that you don't blue screen for IBT after 2 hours does nothing for stability except make sure that it doesn't blue screen during IBT runs. It's so synthetic and out of anything a computer would ever do that it only serves to stress itself. Kind of like Furmark. Not even encoding/folding 24/7 will get you to the same stress levels as IBT.
In the end, I decided to spend a little less money and got the evga 670 GTX FTW instead of the 680 and grabbed the HX750 instead of the HX650. My logic is that I could get another 670 for SLI soon, or replace the card for a better one in the future, and a 750W PSU can last me 5 years if I don't go crazy quad sli.

Thanks everyone for their awesome help, specially mkenyon for his awesome assistance and Hazaro for the builds in the OP.
Not to burst your bubble, but you'll want a 1000W+ PSU for quad videocards generally.

I think 670 over 680 is still a no brainier right now, especially for a single monitor. Good choice!
Hello smart and aesthetically pleasing PC overlords. I'm trying to help my nephew upgrade his PC and struggling a bit on the graphics card.

He needs to run three monitors... so that's the first concern.
Do you mean three monitors for gaming? Or just for general usage, then one monitor for gaming?
 

Sethos

Banned
Finally had to throttle back to 4.5 on my 3930k :( I have this thing where I refuse to operate at over 1.35v ish, it ruins my chi. I just couldn't get it to work straight at those volts, could prime it fine but after 30 minutes of ArmA 2 while watching a movie - Boom, BSOD. No matter how much LLC I added and whatnot it wouldn't stay stable.

Dropped the clock to 4.5 and the voltages, it's rock solid now. Shame.
 

sk3tch

Member
On the flip side, most OC blue screens after a P95 run that occur are during idle, not load. Ensuring that you don't blue screen for IBT after 2 hours does nothing for stability except make sure that it doesn't blue screen during IBT runs. It's so synthetic and out of anything a computer would ever do that it only serves to stress itself. Kind of like Furmark. Not even encoding/folding 24/7 will get you to the same stress levels as IBT.

It is extreme. That's why it's called "Intel Burn Test" - it stresses the components to the maximum. It does everything for stability. Tests temps, tests to make sure there are no freezes, no blue screens, etc. - when your system is completely maxed are the temps within reason? Will it survive the test? If yes, your OC is solid. If no, dial back. It's pretty simple.

EDIT: you can also use the successor that Intel put out: Intel XTU (link in a post above). Similar deal. Synthetic. Just like P95 or any other tool to test stability of an OC.
 

mkenyon

Banned
Right, but the point is that the only thing it's testing for stability in is IBT. Prime 95 on FFT (if no memory overclock) will test your system plenty fine for the most arduous tasks anyone would be doing. Heck, even a 5 hour blend test will have your computer plenty safe for even 24/7 folding.

Ultimately, if you are testing max temps or stability on IBT, it might make you feel more worried than anyone ever should be because it gives artificially high numbers. So what I'm saying, is that it's pretty much pointless if you are testing for a computer you want to OC for normal usage, or even 24/7 folding usage.
 

sk3tch

Member
Right, but the point is that the only thing it's testing for stability in is IBT. Prime 95 on FFT (if no memory overclock) will test your system plenty fine for the most arduous tasks anyone would be doing. Heck, even a 5 hour blend test will have your computer plenty safe for even 24/7 folding.

Ultimately, if you are testing max temps or stability on IBT, it might make you feel more worried than anyone ever should be because it gives artificially high numbers. So what I'm saying, is that it's pretty much pointless if you are testing for a computer you want to OC for normal usage, or even 24/7 folding usage.

Well, to each their own. But I don't worry about the temps as much (due to long term 100% load) - as long as they are within the TJ Max (i.e. 105 F) - my concern is the crashes/hangs/BSODs. Usually within a short period of time you'll experience that. Although as you get closer to the sweet spot you can have it occur a good amount of time into your test.

I guess calling IBT unrealistic is the problem that I have. Yes, it does not mirror real world - but does anything that's benching/stress testing? That's the whole point. To push it to the limit for a specific duration to determine stability. If you feel P95 is fine, cool.
 

Sethos

Banned
According to this guide, http://www.overclock.net/t/1189242/...anations-and-support-for-all-x79-overclockers Prime is the better one for SB / SB-E - Just a random heads up Sk3tch :p

Step #4 Test Stability: For this step you need to monitor conditions, with a program such as CPU-Z for frequency, and then something like HWMonitor, RealTemp, or even motherboard software like EasyTune6-GIGABYTE and TurboV/AI Suite II-ASUS. Then you need a stress testing program, now there are two: Prime95 and IntelBurnTest/LinX. So which is better? Well with Sandy Bridge is it apparent that Prime95 was the better one, even though in the past IntelBurnTest was just as good, it seems that something in the SB uArch is more heavily stressed with Prime95. SB also didn’t have huge issues with heat, one could reach maximum CPU clock without subzero cooling. SBe is not like this, there is no way you can reach your maximum clock on all cores without subzero. So my hunch was that IBT was better, since the heat was going to be more of a killer. So I did a power consumption comparison:

But yeah, I don't give much for these stress tests in real-world results. I could prime my 3930k at 4.7 for hours and hours but 30-40 minutes of ArmA 2 and watching a movie? Boom, BSOD.
 

Ocho

Member
Not to burst your bubble, but you'll want a 1000W+ PSU for quad videocards generally.

That's what I meant. I can do with a 750w for 5 years if I don't find a need to get quad sli (or a crazy single GPU), in which case I would need a new PSU. :p
 

mkenyon

Banned
That's what I meant. I can do with a 750w for 5 years if I don't find a need to get quad sli (or a crazy single GPU), in which case I would need a new PSU. :p
Gotcha!

Yeah, PSU overhead is something I've become more and more a fan of. When I was stuck with two 650W PSUs, it really limited my options in what I could do with hardware. Feels good to have that extra bit of room so you can have an idea and think 'yeah, I can do that'.
 

Mangotron

Member
I have a hard drive/Windows question in two parts.

1. I'm upgrading a X4 635/5770 system to an 8120FX/6950 system, and moving those parts to a system that was a X25000+/9600GT sli system. The hard drives are staying in their original systems, do I NEED to reinstall windows as long as I wipe the graphics drivers? I always have but I'm not sure if it's required.

2. If I do need to do that, I need a fresh partition to install Windows on. Is it possible to make a 12-ish gb partition, install windows, and then merge the partition with my stuff on it and the windows one without losing anything? I couldn't get windows to do this on a system previously and ended up accidentally wiping the entire drive, so if there's a guide for this I'd love to see it.

Thanks for any help in advance!
 
Status
Not open for further replies.
Top Bottom