Support NeoGAF

[oxidax]

If anything your cpu is bottlenecking the 7970...

Don't bother with cross fire until it is fixed. That said, cards don't have to be perfectly matched. As long as they are the same number. 7970 with 7970, 7850 with 7850, etc. manufacturer doesn't matter as far as it cares.

Didn't they fix this back in june with a driver? Thats what i heard.
Or maybe this is something that "mantle" will fix? i hope so..
thanks!

 

[hooijdonk17]

Guys, I have an AM3+ mainboard rated up to 95W. I'm currently using an Athlon II X4 640 3.0GHz CPU with a 1GB Radeon 7770 . Upgrading the GPU is unlikely in the near term, I did that only last year from a 4850 - yeah, I don't like spending huge amounts on hardware. I play on a 16:10 900p monitor at medium-high settings depending on the game.

I noticed that AMD's Vishera CPUs came down in price significantly. Namely the FX-6300 seems to be within my reach. Would it be a worthwhile upgrade?

 

[Zornack]

Is a single 290x going to be enough for 1440p gaming?

I'm looking at BF3 benchmarks and a Titan averages 60 FPS at 2560*1600. That doesn't have me convinced that the 290x, if it ends up being a bit better than a Titan, will be able to run BF4, AC4, WD, Batman, etc. at a smooth 60 FPS with everything maxed out at 2560*1440.

 

[Noctix]

Hi Guys,
Just saw that 7990 dropped to 649 AUD in MSY. Was wondering if it's a card worth getting. I am setting on gtx 295 Need to upgrade asap

 

[tarheel91]

Frame smoothness is just as, if not more important that frame rate. FPS is an out of date tool that is useless in an era of when each second of data will contain 60 (or even 120 for some people) data points. If we were talking about economics you might have a point but we're talking about visual effects, or more specifically how our eyes and brains interpret light. This is nowhere close to the same as taking population, income, or medical stats.

Sure, frame smoothness is important, but did you read what I said? I said FPS is a valuable tool, but you have to understand what it does. It's like peak horsepower for cars. It's a valuable tool for quickly (and roughly) assessing the performance of a car, but ultimately it's but one point on the RPM band among a host of other variables that impact said performance.

Everything you say about FPS applies to these 99th percentile graphs that you guys seem newly obsessed with. They tell you one small thing, and don't necessarily indicate much about smoothness or framerate. If you have one section where there's a sudden dip, 99th percentile frame latency tells you nothing about the experience as a whole. Again, you want the whole story? You look at a histogram and a time graph of frame latency.

Most people don't have the time/inclination to do that, though, so a combination of a few statistical values works well for giving you a good idea of what's going on. Hey, maybe instead of trying to champion a single value (FPS or 99th/whatever percentile) and ignore all others, we could take a traditional statistical approach and use the mean (FPS), median, variance (this is where your percentile stuff comes in), and bias (distribution bias) to look at the distribution of frame latencies. Nah, that makes too much sense and is too inline with traditional analysis in every field ever.

 

[M3z_]

Is a single 290x going to be enough for 1440p gaming?

I'm looking at BF3 benchmarks and a Titan averages 60 FPS at 2560*1600. That doesn't have me convinced that the 290x, if it ends up being a bit better than a Titan, will be able to run BF4, AC4, WD, Batman, etc. at a smooth 60 FPS with everything maxed out at 2560*1440.

Won't know for sure until it's out and benchmarked.

 

[yatesl]

My 4670k BSOD'd after 5 minutes in Prime on 4.3GHz @ 1.25v, but I don't want to increase the voltage as I was already sat at 80c.

Trying 4.2 @ 1.2 as I thought 1.25 was overkill for it. Temps are down to 70c, but I might be being optimistic. We'll see when it blue screens again


Edit: turns out it didn't even BSOD, just straight up shut down and rebooted. I'll try 4.2 @ 1.215v. Am I bad at this, or a bad chip?

 

[AndyBNV]

For everyone concerned about the Watch Dogs specs, they are not final/correct, per a Ubisoft developer.

https://twitter.com/SebViard/status/385815285116063744

 

[kharma45]

BARGAIN ALERT

7970 for £210

http://www.novatech.co.uk/products/...icscards/amdhd7970series/r7970tf3gd5ocbe.html

 

[yatesl]

BARGAIN ALERT

7970 for £210

http://www.novatech.co.uk/products/...icscards/amdhd7970series/r7970tf3gd5ocbe.html

Holy shit. Why can't you SLi a GTX 770 and an AMD 7970...

 

[Hazaro]

So I've got a problem with my PC that I've had for about a year now.

Today we had a few power outages in my area and since getting power back my PC has been shutting down randomly while playing FFXIV. I thought it might be overheating, but temps are normal and I cleaned the case filters/fans.

It just shuts off, then I can't do anything with it for about a minute (it won't power up at all( then it will power on itself.

I'm guessing it's a PSU issue. There were no issues whatsoever before. I've got a Corsair TX750 V2.

Been searching around, and it's not an uncommon problem, guess I'll call Corsair tomorrow.

Well it did eat a lot of power spikes for your PC (My HX520W did as well). I'd give em a call.

I'm pretty bummed. A couple of weeks ago my P8Z77-LX had an overloaded USB port. I had to send it in for RMA. In the meanwhile I bought a shitty little ASRock mATX board to get me through.

The P8 came back to me a few days ago so I put everything back together and couldn't get to post and none of the USB ports were powered. Tried to switch back to the ASRock so I could send the P8 back again and now I can't get to post with it either.

I've tried 1 memory stick, onboard video only, no sound card, clearing cmos, I'm kinda stuck now and have no idea what I should do. I would rather not go out and buy another board.

Anyone have any tips. This is only the 3rd or 4th system I've built so maybe I'm missing something simple. This whole experience this week makes me want to say fuck it and just get a ultrabook and spend my gaming time with Vita + PS4.

The LX is much much worse than the LK. Sorry to hear about your problems. It sounds like you've already tried doing barebones, have you tried using the USB 2.0 ports on both boards?
Have you been shutting off the systems completely (power, wait, unplug/PSU switch/drain power) before mucking around with them?
Memory stick is in the correct 1 DIMM slot? (Check mobo manual)

Lastly try CPU, onboard video, 1 stick RAM outside the case to minimize all other factors and see if it POSTS.

Is a single 290x going to be enough for 1440p gaming?

I'm looking at BF3 benchmarks and a Titan averages 60 FPS at 2560*1600. That doesn't have me convinced that the 290x, if it ends up being a bit better than a Titan, will be able to run BF4, AC4, WD, Batman, etc. at a smooth 60 FPS with everything maxed out at 2560*1440.

WHO KNOWS

My 4670k BSOD'd after 5 minutes in Prime on 4.3GHz @ 1.25v, but I don't want to increase the voltage as I was already sat at 80c.

Trying 4.2 @ 1.2 as I thought 1.25 was overkill for it. Temps are down to 70c, but I might be being optimistic. We'll see when it blue screens again


Edit: turns out it didn't even BSOD, just straight up shut down and rebooted. I'll try 4.2 @ 1.215v. Am I bad at this, or a bad chip?

The heat helps to limit the OC. What board are you using and there are some other settings you can try tweaking to help with your OC, but honestly at 4.2 or 4.3 they shouldn't be needed to be tweaked yet.

 

[OminoMichelin]

Man all these good deals on AMD cards, and in Italy the lowest for a 7950 is 189€(=257$).
Hope we get some deals soon as well.

 

[Chakvr2]

BARGAIN ALERT

7970 for £210

http://www.novatech.co.uk/products/...icscards/amdhd7970series/r7970tf3gd5ocbe.html

Not sure which is better but Scan has an Asus 7970 for £209 with free delivery too:

http://www.scan.co.uk/products/3gb-...hz-gddr5-1010mhz-2048-cores-4x-dp-dvi-d-dvi-i

It dropped from £229 yesterday.

Also would it be worth using my never settle gold coupon now? I know they added Saints Row IV recently but are they likely to add more before December?

 

[yatesl]

The heat helps to limit the OC. What board are you using and there are some other settings you can try tweaking to help with your OC, but honestly at 4.2 or 4.3 they shouldn't be needed to be tweaked yet.

Gigabyte Z87X-UD3H. I can get it running at stable (I think) 4.2 @ 2.5v, but that's obviously not impressive. I'm testing at 4.2 @ 2.3v, and my temps are around 75-80. I might up it and try 4.3 @ 2.7v, but what temps are safe? Air cooled with a 212 EVO

 

[kharma45]

Not sure which is better but Scan has an Asus 7970 for £209 with free delivery too:

http://www.scan.co.uk/products/3gb-...hz-gddr5-1010mhz-2048-cores-4x-dp-dvi-d-dvi-i

It dropped from £229 yesterday.

Also would it be worth using my never settle gold coupon now? I know they added Saints Row IV recently but are they likely to add more before December?

MSI has the better RMA service in the UK so I'd stick with them. BF4 is due at the end of the month according to the guy who works for CCL on here, so I'd wait until then to see.

 

[ekim]

Our local "Saturn" has the following PC as exhibit for 750€:
HP Envy H8
CPU: Intel® Core™ i7-3770
RAM: 16 GB DDR3-RAM
GPU: GeForce GTX 660 (2GB)
HDD: 2 TB (with 16GB SSD Cache -.- )
Power Supply: 460W
Others: Wifi, W8 Pro, Keyboard, Mouse

I know this is a prebuild PC but the price seems ok to me.
The GPU isn't the fastest one but I guess I can swap it out later - or do I also have to switch the power supply then?

Building a similar system at Alternate.de sums up to ~900€
Should I bite for 750€?

 

[yatesl]

My 4670k BSOD'd after 5 minutes in Prime on 4.3GHz @ 1.25v, but I don't want to increase the voltage as I was already sat at 80c.

Trying 4.2 @ 1.2 as I thought 1.25 was overkill for it. Temps are down to 70c, but I might be being optimistic. We'll see when it blue screens again


Edit: turns out it didn't even BSOD, just straight up shut down and rebooted. I'll try 4.2 @ 1.215v. Am I bad at this, or a bad chip?

Following on from this, Prime95 gave me a BSOD on 4.2 @ 1.215v as well. I tried another thing, and that crashed too. Decided to bite the bullet and try 4.3 @ 1.27v. Can I ask, which is the best program? Prime95 shoots my temps RIGHT up, but AIDA and Intel Extreme Tuning has it at different (we're talking Prime95 showing 85c, IXTU showing 65c). Something's obviously not right there. I also noticed that, when doing Prime95, my vcore would increase to almost 1.3v, above what I set in BIOS.

 

[Tal Shiar Agent]

The only game I plan to play in the next year+ on my PC will be Dragon Age: Inquisition. Everything else will be on my PS4. As such I've decided to wait to upgrade my Phenom II x4 965 / 6870 1gb system for a few months.

I want to stick with AMD and I am looking at getting a FX-8350 / 7970 3gb. Will the prices go down on these once the RX200 and whatever their next CPU class is comes out? Or am I better off going with the brand new hardware?

Mind you, I want to spend as little as possible to futureproof my PC for 8th generation games.

 

[yatesl]

The only game I plan to play in the next year+ on my PC will be Dragon Age: Inquisition. Everything else will be on my PS4. As such I've decided to wait to upgrade my Phenom II x4 965 / 6870 1gb system for a few months.

I want to stick with AMD and I am looking at getting a FX-8350 / 7970 3gb. Will the prices go down on these once the RX200 and whatever their next CPU class is comes out? Or am I better off going with the brand new hardware?

Mind you, I want to spend as little as possible to futureproof my PC for 8th generation games.

Prices drop every month - see the 7970 being £100 cheaper than 3 months ago. No doubt there'll be good deals on the R9 cards by the time DA3 is out

 

[batman2million]

Intel 1155 Overclocking Guide

This video helped explain a lot! Thanks! Though I have a 1366.

So following other guides, I'm currently trying to achieve a 3.6ghz but it crashed at 1.1v with in ~5mins using OCCT.

So I upped it to v1.15 left it over night (6 hours about) and when I woke up the computer crashed and wouldn' get off sleep mode. I'm going to try at a high vcore later today but even at 1.15 I noticed that I was reaching temps of 70-75c with in the first 5 minutes.

My highest temp for a core was 82c on a full load. Is that normal? I wanted to try kicking up the voltage to 1.2 to test stability but I'm still concerned about the temps.

I have a cooler master 212 plus with a single fan.

specs:
evga x58 sli
i7 920
12 gb ram @1333 (different sticks, lowest clock is 1333)
corsair 750w psu (something bronze)
gigabyte gtx 570

 

[schandmann]

Our local "Saturn" has the following PC as exhibit for 750€:
HP Envy H8
CPU: Intel® Core™ i7-3770
RAM: 16 GB DDR3-RAM
GPU: GeForce GTX 660 (2GB)
HDD: 2 TB (with 16GB SSD Cache -.- )
Power Supply: 460W
Others: Wifi, W8 Pro, Keyboard, Mouse

I know this is a prebuild PC but the price seems ok to me.
The GPU isn't the fastest one but I guess I can swap it out later - or do I also have to switch the power supply then?

Building a similar system at Alternate.de sums up to ~900€
Should I bite for 750€?

Not a bad deal at all, but upgrading a "mainstream" prebuilt (HP, Dell, etc.) can be a pain.
The PSU will be mediocre and too weak to handle a more power-hungry GPU so get ready to upgrade that along with the GPU. Also, the case might be too small to fit a longer card.
So if you feel you need a new GPU some time down the line, factor in 100-150€ for a new PSU and case. Not worth it, in my opinion.

Spend a little more money and a little bit of your time and build it yourself. Maybe consider buying some parts used if 850€ is too much?

 

[ultimatt]

Sure, frame smoothness is important, but did you read what I said? I said FPS is a valuable tool, but you have to understand what it does. It's like peak horsepower for cars. It's a valuable tool for quickly (and roughly) assessing the performance of a car, but ultimately it's but one point on the RPM band among a host of other variables that impact said performance.

Everything you say about FPS applies to these 99th percentile graphs that you guys seem newly obsessed with. They tell you one small thing, and don't necessarily indicate much about smoothness or framerate. If you have one section where there's a sudden dip, 99th percentile frame latency tells you nothing about the experience as a whole. Again, you want the whole story? You look at a histogram and a time graph of frame latency.

Most people don't have the time/inclination to do that, though, so a combination of a few statistical values works well for giving you a good idea of what's going on. Hey, maybe instead of trying to champion a single value (FPS or 99th/whatever percentile) and ignore all others, we could take a traditional statistical approach and use the mean (FPS), median, variance (this is where your percentile stuff comes in), and bias (distribution bias) to look at the distribution of frame latencies. Nah, that makes too much sense and is too inline with traditional analysis in every field ever.

Probably due to not reading enough on the subject, or having limited exposure to running these types of benches yourself, there is a very key point you are missing out on.

When games aren't having severe issues with drops or big stutters, the 99th percentile figure almost always matches up exactly with Avg FPS. I'm talking within a single digit matching, or even right on the money.

In this respect, your assessment of why Avg FPS is a good tool to roughly estimate performance not only equally applies to 99th Percentile, but in fact becomes surpassed by 99th Percentile. The reason being that 99th Percentile takes into account those issues with stuttering.

Yes, the full list of every single frame during the course of the benchmark is going to be the absolute best for determining performance and getting an accurate idea of what is going on. But, you pointing that out as part of your FPS spiel is nearly meaningless. It has nothing at all to do with Avg FPS vs. 99th Percentile. I'm not sure why it's even being discussed. Yes, the sky is blue.

 

[spindoctor]

I'm looking to do a partial upgrade of my system and these are the parts I had mind -

i5 4670k
Asus Z87-A mobo (Amazon link)
Gskill 8gb (2x4gb) ddr3 1600 ram (Amazon link)
Cooler Master Hyper 412 Slim cpu cooler (Amazon link)

These components will pair with my existing GTX670, Corsair 750W PSU, hard drives etc. in my CoolerMaster CM690 case.

Was wondering if anyone had any thoughts about this config. If there's anything else I should look at or maybe if there are potential issues. I'd appreciate any feedback I'd get.

 

[opticalmace]

Is forcing AA (through e.g. CCC) worse than letting the application decide?

As in, is there extra overhead or complications because of it?

 

[Lostconfused]

I am starting to feel like I need to upgrade from an i5 750.

Currently thinking of 4820K over 4770K for potentially better overclocking.

 

[Bucket-o-roadkill]

If I'm getting a 4670k, is a Cooler Master Hyper TX3 EVO any good? It's a tenner cheaper than the 212 evo......

 

[kennah]

If I'm getting a 4670k, is a Cooler Master Hyper TX3 EVO any good? It's a tenner cheaper than the 212 evo......

There's a reason for that. Spend the $10.

 

[Bucket-o-roadkill]

There's a reason for that. Spend the $10.

Haha, ok fair shout. I need to get out of my scrimpy ways when buying this machine.. I was close to ordering, now I'm getting the 212. Thanks my man. While I'm here, how is the MSI Z87-G43? I've already ordered that as I saw it for under 75, madness.

Also MSI 7950 3gb boost iceq, decent card?

 

[Seanspeed]

Intel 1155 Overclocking Guide

I'm about halfway through this video and I think I'm just not comfortable overclocking now. I was thinking it would be much simpler than this, but he's changing all sorts of settings and I'm clueless about what a lot of this stuff means like vCore loadline calibration and whatnot. I actually don't quite get how voltages work or what 'stable' means or anything, either.

I also don't understand how long you're supposed to test stuff. He's doing it for 30 minutes, but I hear other people say you've gotta do it for like 24 hours.

 

[Felix Lighter]

I'm about halfway through this video and I think I'm just not comfortable overclocking now. I was thinking it would be much simpler than this, but he's changing all sorts of settings and I'm clueless about what a lot of this stuff means like vCore loadline calibration and whatnot. I actually don't quite get how voltages work or what 'stable' means or anything, either.

I also don't understand how long you're supposed to test stuff. He's doing it for 30 minutes, but I hear other people say you've gotta do it for like 24 hours.

While you are making adjustments, you can run stress tests quickly because if you are not close to stable it will usually crash quickly, if it even boots in the first place. When you finally feel you have hit a stable overclock (no blue screens or sudden crashes) and optimized your voltage and are comfortable with the max temperature you run the stress test for 16-24hours to be sure that it's completely stable for 24/7 usage. If it crashes, you either have to increase voltage or reduce your overclock and try it again. It's really not as overwhelming as it can initially seem.

 

[Seanspeed]

While you are making adjustments, you can run stress tests quickly because if you are not close to stable it will usually crash quickly, if it even boots in the first place. When you finally feel you have hit a stable overclock (no blue screens or sudden crashes) and optimized your voltage and are comfortable with the max temperature you run the stress test for 16-24hours to be sure that it's completely stable for 24/7 usage. If it crashes, you either have to increase voltage or reduce your overclock and try it again. It's really not as overwhelming as it can initially seem.

Yup, thanks. I watched the rest of the video and the voltage thing was making more sense. I never realized that you want to minimize voltages and why that's so important. So I take it up step-by-step with quick tests and then a long one once I've found the performance I'm looking for?

Can I really just ignore all that other stuff he mentions about vCore load calibration stuff and whatnot?

And what about Turbo? I don't quite understand how that factors into overclocking. If I'm at 4.2Ghz, is it going to speed up higher than that? Does overclocking mean that I'm always running at 4.2Ghz(since right now it sits at 1.6Ghz when not doing anything). My voltage also seems to be variable when I look at CPU-Z with stock speeds. Does that lock down after overclocking? Is it going to be jumping around based on load?

 

[Parsnip]

Well, I'm currently running a OCCT large data set test after OC'ing my i5-750 to 3.6ghz(35% OC), it has been running almost an hour now, temps are hovering between 65C and 70C. Seems stable I guess. I assume that OCCT will tell me immediately if it errors or something.

Should I just move to the small data set to generate more heat or what?

 

[Felix Lighter]

Awwww never mind.

 

[Kintaro]

I am really interested in that fRactal Design Node build in the OP, but I can't decide which way to go. Ivy or Haswell. ;_;

 

[LCGeek]

Yup, thanks. I watched the rest of the video and the voltage thing was making more sense. I never realized that you want to minimize voltages and why that's so important. So I take it up step-by-step with quick tests and then a long one once I've found the performance I'm looking for?

Can I really just ignore all that other stuff he mentions about vCore load calibration stuff and whatnot?

And what about Turbo? I don't quite understand how that factors into overclocking. If I'm at 4.2Ghz, is it going to speed up higher than that? Does overclocking mean that I'm always running at 4.2Ghz(since right now it sits at 1.6Ghz when not doing anything). My voltage also seems to be variable when I look at CPU-Z with stock speeds. Does that lock down after overclocking? Is it going to be jumping around based on load?

Well in the case of my 3570 (non-k)

without turbo the max multiplers I can use is 34x with the feature on it's 40x four all cores. For record you need a board with the multicore feature to apply turbo past a certain point for all cores. Turbo will not let you got any more speed than what you're setting. Rather if you don't need turbo don't use it since people say it can play with stability especially in high load situations.

Are you using offset or fixed voltage. Until you turn off speed step and C1 and C6 features your cpu will just in voltage, which isn't a bad thing considering it saves on power. When you use something that needs it the cpu will jump to the right voltage settings and go back down afterwards.

 

[muu]

Yup, thanks. I watched the rest of the video and the voltage thing was making more sense. I never realized that you want to minimize voltages and why that's so important. So I take it up step-by-step with quick tests and then a long one once I've found the performance I'm looking for?

Can I really just ignore all that other stuff he mentions about vCore load calibration stuff and whatnot?

And what about Turbo? I don't quite understand how that factors into overclocking. If I'm at 4.2Ghz, is it going to speed up higher than that? Does overclocking mean that I'm always running at 4.2Ghz(since right now it sits at 1.6Ghz when not doing anything). My voltage also seems to be variable when I look at CPU-Z with stock speeds. Does that lock down after overclocking? Is it going to be jumping around based on load?

With speedstep enabled you'll still downclock when idle.

 

[Crisco]

Well, I'm currently running a OCCT large data set test after OC'ing my i5-750 to 3.6ghz(35% OC), it has been running almost an hour now, temps are hovering between 65C and 70C. Seems stable I guess. I assume that OCCT will tell me immediately if it errors or something.

Should I just move to the small data set to generate more heat or what?

It's probably good at that speed. Those are fairly low temps too, so you probably have a good bit more thermal headroom. I'd shoot for ~3.8ghz and then run Prime95 overnight.

 

[Seanspeed]

Well in the case of my 3570 (non-k)

without turbo the max multiplers I can use is 34x with the feature on it's 40x four all cores. For record you need a board with the multicore feature to apply turbo past a certain point for all cores. Turbo will not let you got any more speed than what you're setting. Rather if you don't need turbo don't use it since people say it can play with stability especially in high load situations.

How do I know if I need Turbo? I don't think I quite understand what it is in the first place. I'm just scared its going to mess something up after I overclock.

Are you using offset or fixed voltage. Until you turn off speed step and C1 and C6 features your cpu will just in voltage, which isn't a bad thing considering it saves on power. When you use something that needs it the cpu will jump to the right voltage settings and go back down afterwards.

I don't know what the difference is between offset or fixed voltage. I didn't realize there were different types. And I think you skipped a word or something with the bolded part.

The guy in the video said something about an 'auto' setting that can make your voltages unnecessarily high. And what about it going too low? If I turn off auto, does it just stay at what I put it as no matter what?

I really don't know what I'm doing or what any of this stuff means. I was kinda hoping there'd be a simple 'do x and then y' sort of thing, but it seems you should really understand what you're doing before jumping in. Not sure I'm comfortable with that.

With speedstep enabled you'll still downclock when idle.

Is it ok to keep that on? Is there any reason I shouldn't?

 

[jim2point0]

I also don't understand how long you're supposed to test stuff. He's doing it for 30 minutes, but I hear other people say you've gotta do it for like 24 hours.

I've overclocked my 3770k to 4.4 Ghz.

I don't think I've ever let my tests run longer than 10 minutes. In my experience, if it doesn't crash in that time, it's pretty stable. But I also don't try to push my stuff too hard. Some folks go much higher than that with this CPU.

I'm glad my motherboard (Asus Maximus V Extreme) makes this relatively easy. I set a few things to "enable," turn my mutliplier up to 44, and change my voltage. That's it. Then I test for a while. If it's too hot, I lower voltage. If it's unstable, I raise voltage. Eventually I come up with a configuration I'm happy with.

 

[yatesl]

Is there any reason why Prime95 is pushing my vcore to 1.248v, when I have it explicitly set to 1.22v in BIOS?

CPU-Z reads it at 0.780v whilst idle (although fluctuates), but as soon as I put it under load it shoots past my set vcore setting. It's only running at 4.2GHz (i5 4670K), and at that vcore it's kicking out temps of 75c. At a relatively low overclock, I'd rather have the vcore lower.

Is there a setting I'm missing? It's a new Gigabyte motherboard.

Intel Extreme Tuning Utility pushes it to 1.236v - only at 63c, but still more than I wanted. I'm obviously not running at load 100% (only when playing games perhaps?), but I'd rather not have those red hot temps if I can help it.

 

[LCGeek]

How do I know if I need Turbo? I don't think I quite understand what it is in the first place. I'm just scared its going to mess something up after I overclock.

Do you have a K processor, which is unlocked. If you do, do not use turbo imo it's unnecessary. Also it won't mess you up I'm using turbo right now.

I don't know what the difference is between offset or fixed voltage. I didn't realize there were different types. And I think you skipped a word or something with the bolded part.

I did sorry it happens when I type to fast. I want to say the cpu will just up the voltage when needed and then back down when you're not using a heavy load.

In the case off offset vs fixed. A fixed voltage only has one setting, offsetting combined with speedstep in the bios will allow the cpu to go between various states of voltage.

The guy in the video said something about an 'auto' setting that can make your voltages unnecessarily high. And what about it going too low? If I turn off auto, does it just stay at what I put it as no matter what?

I really don't know what I'm doing or what any of this stuff means. I was kinda hoping there'd be a simple 'do x and then y' sort of thing, but it seems you should really understand what you're doing before jumping in. Not sure I'm comfortable with that.

Auto offset can add to much though in my experience in this case it was only a few more points than the fixed value I used. If you set the offset you can have it use as much as it needs for a game and then for browsing it will use only a necessary amount. When you idle it would be at it lowest.

Is it ok to keep that on? Is there any reason I shouldn't?

Perfectly fine the only issue it adds to is dpc latency which for most is rarely an issue. Only time some turn it off is like me when you want a fixed voltage to see what the max overclock on your cpu is. With it turned off along with C1 and C6 you can keep the fixed voltage at all times which is good turning benchmarking or error testing. Once you're done enabling these again is perfectly fine and almost better long term if you want to save on power which can extend the life of your cpu.

 

[Tonezorz]

I haven't played more with overclocking due to all these guides being for different motherboards than I have. I'm running the MSI Z87 G45 Gaming, which seems to have many additional / different settings than the Gigabyte / Asus boards used in the good guides.

I get spooked and think I'm going to mess it up. :/

 

[Parsnip]

It's probably good at that speed. Those are fairly low temps too, so you probably have a good bit more thermal headroom. I'd shoot for ~3.8ghz and then run Prime95 overnight.

Sounds good. OCCT defaulted to 85C as its warning level, do you know if that's a common temperature or did it fetch that from somewhere based on my cpu? Really just curious what kind of headroom I do have.


Edit: Also, should I be worried about my PSU? It's about 5 years old Corsair HX520, I'm not going to make it blow up or anything, right?
Rest of my system is 3 hard drives, Asus P7P55D PRO mobo, Gigabyte GTX 760 and 2x4gigs of some RAM which speed and brand I'm completely blanking on right now for some reason.

 

[Xyber]

So I'm thinking of upgrading to 1150 MoBo and an i7 4770K. And when I re-install my PC I want to use my Win 8 upgrade licence, but I understand that you must transfer it somehow to the new MoBo?

I've seen some stuff about doing an upgrade again on the new install (Win 7 first and then use the upgrade assistant to upgrade to 8), will that work?

So how would I go about to get Win 8 on this new install so that it activates properly?

 

[Spiritreaver]

So I seemed to have fixed my random shutdowns I was having last night. I updated my BIOS, seems it was well over 2 years old. So far so good, PC's been running like a champ for the past 3 hours.

If it fails again, the replacement PSU is on it's way, but I'll probably just end up shipping it back.

 

[kennah]

So I'm thinking of upgrading to 1150 MoBo and an i7 4770K. And when I re-install my PC I want to use my Win 8 upgrade licence, but I understand that you must transfer it somehow to the new MoBo?

I've seen some stuff about doing an upgrade again on the new install (Win 7 first and then use the upgrade assistant to upgrade to 8), will that work?

So how would I go about to get Win 8 on this new install so that it activates properly?

You just install it. It should authenticate fine. If it doesn't you just call the phone number and tell them you installed a new motherboard. Takes 10 minutes.

 

[Crisco]

Sounds good. OCCT defaulted to 85C as its warning level, do you know if that's a common temperature or did it fetch that from somewhere based on my cpu? Really just curious what kind of headroom I do have.


Edit: Also, should I be worried about my PSU? It's about 5 years old Corsair HX520, I'm not going to make it blow up or anything, right?
Rest of my system is 3 hard drives, Asus P7P55D PRO mobo, Gigabyte GTX 760 and 2x4gigs of some RAM which speed and brand I'm completely blanking on right now for some reason.

85C is the generally accepted max temperature for Nehalem chips. Realistically you'll never see it get that high during normal use, only stress testing. Still, for maximum longevity, try and keep it under 80C.

 

[tarheel91]

Probably due to not reading enough on the subject, or having limited exposure to running these types of benches yourself, there is a very key point you are missing out on.

When games aren't having severe issues with drops or big stutters, the 99th percentile figure almost always matches up exactly with Avg FPS. I'm talking within a single digit matching, or even right on the money.

In this respect, your assessment of why Avg FPS is a good tool to roughly estimate performance not only equally applies to 99th Percentile, but in fact becomes surpassed by 99th Percentile. The reason being that 99th Percentile takes into account those issues with stuttering.

Yes, the full list of every single frame during the course of the benchmark is going to be the absolute best for determining performance and getting an accurate idea of what is going on. But, you pointing that out as part of your FPS spiel is nearly meaningless. It has nothing at all to do with Avg FPS vs. 99th Percentile. I'm not sure why it's even being discussed. Yes, the sky is blue.

I'm gonna call bullshit on that, sorry.

http://www.anandtech.com/show/6994/nvidia-geforce-gtx-770-review/7

95th percentile seems to be ~10% less than average frame rate. 95th percentile is 1.65 standard deviations. 99th percentile is 2.33 standard deviations. Assuming a normal distribution, this says 99th percentile will be ~14% less than average. The fact of the matter is that how representative the 99th percentile is of the population as a whole will depend on how consistent the frame rate is. For more taxing games, the delta will be even greater (for Crysis 3 in the same article, the 95th percentile is ~25% off the average, so the 99th would be ~35%). That's the difference between 30 and 22FPS, or 40 and 30.

 

[Xyber]

You just install it. It should authenticate fine. If it doesn't you just call the phone number and tell them you installed a new motherboard. Takes 10 minutes.

Alright, thanks.

 

[Erebus]

Is there any reason why Prime95 is pushing my vcore to 1.248v, when I have it explicitly set to 1.22v in BIOS?

CPU-Z reads it at 0.780v whilst idle (although fluctuates), but as soon as I put it under load it shoots past my set vcore setting. It's only running at 4.2GHz (i5 4670K), and at that vcore it's kicking out temps of 75c. At a relatively low overclock, I'd rather have the vcore lower.

Is there a setting I'm missing? It's a new Gigabyte motherboard.

Intel Extreme Tuning Utility pushes it to 1.236v - only at 63c, but still more than I wanted. I'm obviously not running at load 100% (only when playing games perhaps?), but I'd rather not have those red hot temps if I can help it.

I'm not familiar with your motherboard but I assume you have left some option somewhere set on Auto perhaps? Do you use offset voltage or fixed set to 1.22V?


So fellow PC-GAF, I need some help troubleshooting my PC woes but first let me recap a few things.

If you check my last couple of posts in this thread, you'll see that I recently had trouble with my old rig. It would randomly BSOD'd or become completely unresponsive to any mouse or k/b input while idling. After running some tests, I blamed these issues on the RAM (RAM errored like crazy during memtest). Since my old rig was based on LGA 775, I saw that as an opportunity to finally upgrade to an Ivy Bridge CPU. Obviously, I bought new motherboard and RAM along with the CPU and from the old system I only kept the GPU (a GTX 460 1GB) and the PSU (which is practically new as I replaced my old one back in spring).

I did a fresh install of Windows 7 and everything was going smoothly until a moment ago that I decided to play L4D2. As soon as the game loaded, I noticed that I was getting half the FPS. The game itself is not that demanding thus I was always getting stable 60FPS with everything on max vsynced. Now the game was running like a slideshow staying under the 30FPS mark for no reason, like something was throttling the GPU. I quit back to windows, launched again the game but the same thing happened. When I quit to windows for the second time, I noticed that the mouse cursor was freezing every 2-3 seconds. I did a system restart, booted back to Windows, run L4D2 again and guess what? The game was running normal at 60FPS, no slowdown. At this point I should mention that before the restart, I had Firefox running in the background which as of lately was causing weird issues with the graphics card, constantly crashing the nVidia drivers or the whole system (referring to the old LGA 775 system) until I turned off the hardware acceleration from the settings.

I'm wondering could this be just Firefox affecting the graphics drivers or is it a hardware problem? Could a dying GPU behave like that? Like I said almost everything on my PC right now is new except for the GPU, PSU and well the HDD.


UPDATE:

OK it happened again. Mouse cursor stalling/freezing every couple of seconds and I was just surfing the internet on Firefox! While this was happening, I noticed this:

GPU frequency stays at 725MHz with no load whatsoever (normally it should be around 50Mhz when idle) but paradoxically the memory was underclocked like it should be under the circumstances.

Digging even further I found this:

https://forums.geforce.com/default/...hread-released-8-20-13-/post/3895502/#3895502

https://forums.geforce.com/default/...hread-released-8-20-13-/post/3895565/#3895565