775 system: Upgrade or total refit?

[lehtv]

I get what your saying but if you use your PC primarily for gaming and don't do compression and video encoding then there really is really no benefit going above 1333MHz IMO

I agree, Anandtech's memory scaling comparison confirms that faster RAM is absolutely useless for gaming on SB.

also is there a time when you should ever recommend a board over $200? even if he has a big budget it's still money down the drain.

Yes, I could recommend a $200 board to someone who has very high quality and connectivity requirements and a big budget. That's obviously not the only variable though. If you pay $300 for a 2600K or even more for the upcoming SB-E, $60 for a high quality CPU cooler, you only need to spend another $140 on a motherboard to reach $500. Adding 16GB 1333MHz RAM worth $90 will make that a $590 setup. 16GB 1600MHz is only $10 more, increasing the price of your core setup by ~1.6% which is comparable to the performance increase you get in encoding applications. A small increase in price for a small increase in performance - I'm not going to complain if someone makes that choice.

Well, he gives baffling advice like how a 955BE + 6950 2GB is somehow better than a 2500K + 560 Ti 1GB for a $700 build, and now this. I don't even know whether if he is actually using his head to think.

1) That's off topic. If you want to discuss that "baffling advice" why don't you discuss it in the actual thread were the advice was given?

2) You're misrepresenting my position on that matter. There's a lot more to it than "2500K + 560 Ti versus 965BE + 6950 2GB". If you had read my posts in that thread, you'd know this. The 2500K build sacrifices a lot of things that I would prefer not to sacrifice (YMMV) unless the idea is to build an $800 rig, not a $700 rig.

 

[mfenn]

Alrighty then. I changed the order but ended up with the Corsair HX650 instead. The modular thing does appeal to me after all...

Thanks for all your help!

No prob, glad you stuck with it!

I don't think that modular is really worth it because the cables tend to either (a) get used inside the case anyway or (b) take up space in a drawer/get lost. Better to keep everything inside the case where it is easily accessible when needed and out of sight when not IMHO.

 

[blacktankofhopelessness]

I see your point. But since I'm a bit on the obsessive-compulsive side when it comes to the order of things I think modular is the way to go. All that clutter of cables has annoyed me more than once...

 

[lehtv]

(b) take up space in a drawer/get lost.

When I buy hardware I keep the packaging anyway. Not just to store extra cables and documentation, but also in case I want to store the component away after an upgrade or sell it forward.

 

[mfenn]

I see your point. But since I'm a bit on the obsessive-compulsive side when it comes to the order of things I think modular is the way to go. All that clutter of cables has annoyed me more than once...

Fair enough.

When I buy hardware I keep the packaging anyway. Not just to store extra cables and documentation, but also in case I want to store the component away after an upgrade or sell it forward.

You must have a bigger place than I! :awe: I have hardware spilling out of my closets before even worrying about the packaging.

 

[blacktankofhopelessness]

Ok, so I finally got myself a new rig!

As you can see in the sig I also decided to get me a pretty little SSD for my OS and games.

I ran into some serious BSOD-issues at first, with Windows dying randomly after about 30 or so minutes of use. But I think I managed to get stable after identifying the SSD as the main culprit. I sorted it out by doing the following:

1. Flashed mobo BIOS
2. Set Mem voltage to manual and 1.5 volts
3. Ran Memtest 86+ (excluding RAM as the culprit)
4. Updated SSD Firmware
5. Disabled HD-shutdown in Windows power management and moved all temp folders to my HDD. Also disabled Superfetch, Windows Search and Disk defrag.

The first three steps didn't help - I got advice from the official Corsair forums telling me to do that before updating the SSD firmware. The fourth step however seems to have gotten my system stable.

I used my sys in every way possible after that fourth step - ran Prime 95, ATTO, surfed the web, and did about an hour of gaming, including letting my rig go into sleep mode a few times just to see if it would wake up again (this was one of the occasions it would otherwise go BSOD on me before). So far no new issues.

I'm hoping all is well now, because I really love my new rig and especially the SSD which makes everything, and I mean everything run so smooth and quick. Great investment... as long as it works!

Any thoughts or other advice on what I can do to make my system as stable as possible? I'm not overclocking at the moment. That can wait.

 

[lehtv]

Nice

Any thoughts or other advice on what I can do to make my system as stable as possible? I'm not overclocking at the moment. That can wait.

For when you do OC: http://forums.anandtech.com/showthread.php?t=2195063

If you get any BSODs in the future you can go to http://www.sevenforums.com/crashes-debugging/96879-blue-screen-death-bsod-posting-instructions.html, they'll analyze BSOD mem dump files for you

 

[blacktankofhopelessness]

Nice

For when you do OC: http://forums.anandtech.com/showthread.php?t=2195063

If you get any BSODs in the future you can go to http://www.sevenforums.com/crashes-debugging/96879-blue-screen-death-bsod-posting-instructions.html, they'll analyze BSOD mem dump files for you

Ok, that's good to know.

I've managed to open and read dump files but can't make much sense of them other than the fact that it was csrss.exe that always crashed causing the BSOD. f4 was the base error code. Hopefully that issue won't return, but if it does I now know where to turn for help, so thanks!

 

[mfenn]

Glad you got it working! If your Force 3 that was an old stock from the original batch, I'm not surprised that you had firmware issues. You can probably move your temp files back the SSD now that you've got it stable.

 

[blacktankofhopelessness]

Glad you got it working! If your Force 3 that was an old stock from the original batch, I'm not surprised that you had firmware issues. You can probably move your temp files back the SSD now that you've got it stable.

You mean the line of SSD's that Corsair recalled? According to Corsair only the 120GB version of the Force 3 series with serial numbers starting with 1123 and below were affected. Mine starts with 1330-something.

But yes, a lot of people seem to have had issues with the Force 3 series in general, at least judging by the number of posts in the f4 BSOD-thread over at the Corsair tech support forums. Mine was running the 1.3 firmware version before I updated to 1.32.

Still no issues by the way. Going to try overclocking soon!

EDIT: A moderate overclock of 100x40 = 4.0 Ghz was simple enough and I am stable at just 1.25 Vcore. Nice!

 

[mfenn]

Glad to hear on the SSD.

Bump that multi to 45 and get back to us. Then you'll be cooking with gas!

 

[blacktankofhopelessness]

I'm going to try 45 tonight. But I don't want to produce too much excessive heat.

When I went from 3.3 Ghz to 4 my CPU load temp during Prime95 rose from around 53C to 64C (ambient temp around 24C). Does that sound normal, especially given that I should have a pretty good cooler? Granted, I'm running the Scythe Ninja III at 800-900 rpm so as to avoid too much noise. Tcase for the i5-2500K is 72C so I don't want to get too close to that number.

Will I get much better gaming performance with 4.5Ghz as opposed to 4? If not, I'm not sure it's worth the additional heat/noise for me.
----------------------------

By the way, I got a Nexus 120mm case exhaust fan that is supposed to be the best near-silent fan according to Silentpcreview. It's rated at 18db but must run much higher because it is now the noisiest fan in my rig (it runs at ~1100 rpm). All of the other fans run very quietly, including my near-silent GTX460. Before installing the case fan you couldn't tell there was a computer in the room, now you can if you listen closely.

I'm thinking I should undervolt it to 7v to go near-silent but I'm unsure how to do that safely. Could anyone point me in the right direction? And yes, the fan lacks a Qfan-controller, or a fourth cable if you will, so there's no way of controlling the fan through BIOS settings or Speedfan.

EDIT: Ordered a Zalman fan mate 2 (5$), so that takes care of question nr 2!

 

[lehtv]

That sounds pretty normal, yes. Antec Sonata III isn't great for CPU cooling, a more cooling-oriented case would improve CPU temps by several degrees.

What are you measuring Tcase with?

Will I get much better gaming performance with 4.5Ghz as opposed to 4?

Well.. (4.5/4) * 100% = 12.5% faster in CPU intensive tasks. In games, much less. I'd expect under 5% better frame rates. It's also probably true that every increase in clock frequency above that 4GHz point will only consume more power and put out more heat per MHz. 4Ghz seems to be the MHz/watt sweet spot (based on data here). If I had a 2500K, I'd OC it to 4.0 and keep the CPU cooler in silent mode.

I'm thinking I should undervolt it to 7v to go near-silent but I'm unsure how to do that safely.

http://www.newegg.com/Product/Product.aspx?Item=N82E16811996016 would work too

And yes, the fan lacks a Qfan-controller, or a fourth cable if you will, so there's no way of controlling the fan through BIOS settings or Speedfan.

Err, I think the BIOS Qfan setting just adjusts the voltage supplied to the fan. The fourth cable is for PWM which is used to adjust RPM (voltage?) on the fly so that the fan will spin up at higher load as determined by the motherboard. Someone correct me if I'm wrong; but in my case, 3-pin fans run at lower speed if set to silent mode by Qfan.

 

[sm625]

There is something about a quadcore that just cant be quantified by benchmarks alone. I have a program that loads down a cpu with a very light load, and I run many instances to produce the load I want. On my 3.2GHz E6600 here I can only run about 4 instances before I can really feel my system becoming laggy. I feel it even with the cpu only at 50%.

Now on my 3.1GHz Q6600, I can run 12 instances of my program and barely even notice it. When I do get a lag spike it is much shorter.

Since a Q6600 is just two E6600s, I would expect to see the same type of unresponsiveness at 8 instances compared to the E6600 at 4 instances. But I dont.

 

[blacktankofhopelessness]

@lehtv: Well, I'm going by Realtemp to measure temps. Nevermind Tcase, I have no way of measuring that.

Sorry for making you refer to that thread again. You had already pointed me in that direction previously. Laziness I guess. Anyway, I think I'll stay at 4 Ghz to keep my rig quiet.

That cable adapter looks good, but the fan mate I just ordered was even cheaper and lets me adjust fan speed without disconnecting/connecting cables.

About that fourth cable - are you sure? I've never gotten changed Qfan settings to work with three pinned fans. My Nexus fan just buzzes away at max speed regardless of which setting I put it in (BIOS Qfan settings). I also tried setting the max fan rotation to 50% but that didn't work either.

 

[lehtv]

About that fourth cable - are you sure? I've never gotten changed Qfan settings to work with three pinned fans. My Nexus fan just buzzes away at max speed regardless of which setting I put it in (BIOS Qfan settings). I also tried setting the max fan rotation to 50% but that didn't work either.

Yes I'm pretty sure. Otherwise why would the motherboard even offer Q-fan control on its 3-pin fan headers?

 

[blacktankofhopelessness]

Yes I'm pretty sure. Otherwise why would the motherboard even offer Q-fan control on its 3-pin fan headers?

Nevermind me, I'm an idiot. An idiot. I had connected the 3-pin fan header to the mobo AND the fourpin IDE power connector to the PSU. Brilliant piece of thinking there on my part. No wonder the Qfan settings didn't work...

All set now and yes, the Qfan settings work as they should.

Ok, you can all go ahead and laugh now.
----------------

But I don't care! My rig is near-silent!

 

[lehtv]

But I don't care! My rig is near-silent!

At least you can hear us laughing *hahahaha*

 

[blacktankofhopelessness]

Oh bollocks. Wait, let me just ramp up that fan speed and put on some headphones...

 

[mfenn]

Well.. (4.5/4) * 100% = 12.5% faster in CPU intensive tasks. In games, much less. I'd expect under 5% better frame rates. It's also probably true that every increase in clock frequency above that 4GHz point will only consume more power and put out more heat per MHz. 4Ghz seems to be the MHz/watt sweet spot (based on data here). If I had a 2500K, I'd OC it to 4.0 and keep the CPU cooler in silent mode.

My thinking on that topic is that the i5 2500K turbos to 3.7 to begin with, so why even mess with overclocking at all if you're just going to 4?

Err, I think the BIOS Qfan setting just adjusts the voltage supplied to the fan. The fourth cable is for PWM which is used to adjust RPM (voltage?) on the fly so that the fan will spin up at higher load as determined by the motherboard. Someone correct me if I'm wrong; but in my case, 3-pin fans run at lower speed if set to silent mode by Qfan.

There are two main ways to control the speed of a fan in a PC, voltage and pulse width modulation (PWM). A 3-pin fan can only use voltage, whereas the 4th pin on a 4-pin fan is for PWM control.

Grossly simplified, a brushless DC fan like a PC fan spins the blades by using an electromagnetic field to exert an angular force on the hub of the fan. A simple way to think of the difference between voltage and PWM is that altering the voltage changes the strength of the push whereas PWM changes the length of the push. The disadvantage of voltage control is that a certain amount of force is required to overcome the friction in a fan, and as such there is a minimum voltage required to start the fan and also to keep the fan moving. PWM doesn't have that problem and can provide a much wider range of control.

 

[lehtv]

My thinking on that topic is that the i5 2500K turbos to 3.7 to begin with, so why even mess with overclocking at all if you're just going to 4?

I don't understand that logic. Turbo boost just means 3.7Ghz effective. Overclocking to 4Ghz means 4Ghz effective, assuming you disable turbo boost. Either way, the CPU will idle at low clocks. And 4 is better than 3.7, so why not OC?

Also, the amount of turbo boost depends on the number of cores active (with all four cores you don't get 3.7, but only 3.4 or 3.5, not sure), while non-turbo OC is the same for every scenario.

Thanks for the PWM explanation :thumbsup:. It sounds a bit like in physics where a wave's energy can be increased or decreased by two ways: changin the wave amplitude, or changing the frequency (and as a result, wave length).

 

[mfenn]

I don't understand that logic. Turbo boost just means 3.7Ghz effective. Overclocking to 4Ghz means 4Ghz effective, assuming you disable turbo boost. Either way, the CPU will idle at low clocks. And 4 is better than 3.7, so why not OC?

Also, the amount of turbo boost depends on the number of cores active (with all four cores you don't get 3.7, but only 3.4 or 3.5, not sure), while non-turbo OC is the same for every scenario.

Cause 300MHz will hardly be noticed and changing that one number in the BIOS is a lot of effort. :awe:

Thanks for the PWM explanation :thumbsup:. It sounds a bit like in physics where a wave's energy can be increased or decreased by two ways: changin the wave amplitude, or changing the frequency (and as a result, wave length).

You got it, that's very analogous to what's happening with PWM vs. voltage control.

 

[blacktankofhopelessness]

Crap.

This morning I had the best and quietest system I have ever built. Now I have a piece of junk that refuses to even post.

I was happily watching video previews of this and that when all of a sudden my rig went completely dead. Tried to reboot, nothing. Disconnected the power chord, cleared CMOS and removed the battery. Pushed the power button and yes, the fans start spinning but nothing else. Not a single peep and my screen is still blank.

I proceed to unplug everything except the main ATX power plug, PSU connector, RAM, and CPU fan. Same result. Cleared CMOS without battery for 15 mins. No change. Tried booting without RAM, nothing. Tried booting without RAM but with the GPU plugged in. Still nothing. Not a single beep or indication as to what might be wrong.

I suspect the mobo so have left the entire rig for checking by a hardware pro. Then when I'm sure RMA whatever isn't working.

Crap.

Edit: I was running my rig on a mild OC of 4.0 Ghz. (Load-Line Calibration: Optimized, Vcore 1.225, CPU PLL 1.8, VCCSA auto, VCCIO auto, RAM voltage 1.5)

Never had any temp issues: 62C max under load and 32C Idle.

What do you think? Mobo or maybe CPU?

 

[mfenn]

Most likely mobo, second guess would be PSU. If you don't have spare parts laying around, taking it to a shop or something where they can test it is probably your best idea.

 

[blacktankofhopelessness]

I sold my previous components already, so no, no spare parts. First time I've had this many problems with a new build.

*sigh*