or Connect
AVS › AVS Forum › Video Components › Home Theater Computers › Mfusick's How to build an affordable 30TB Flexraid media server: Information Requested.!
New Posts  All Forums:Forum Nav:

Mfusick's How to build an affordable 30TB Flexraid media server: Information Requested.! - Page 73

post #2161 of 3346
Thread Starter 
Quote:
Originally Posted by rc05 View Post

I use tRAID for general non-media storage. Performance is acceptable with OS caching turned on. I only have 2 DRU + 1 PPU right now though.

What is acceptable ? 80MB/sec ???
post #2162 of 3346
Quote:
Originally Posted by Mfusick View Post

What is acceptable ? 80MB/sec ???
Wait, too much is almost enough? Obviously nothing less than fully saturating a 10Gig LAN is enough smile.gif

Noticed you mentioned upgrading the G860 from what I understand must be a second server?

I'm upgrading and dedicating my server soon as well biggrin.gif

Bought a used 3770k off Amazon from somebody for $190 ?? Seemed like a fair deal to me, and a worthwhile gamble. Seller said they've been using it for a year with a Corsair Hydro H50, so hopefully it's never been in dangerous temp territory.

I'm going to actually dedicate this server/htpc to server duty. Even if I have to cut holes in walls smile.gif I'll pick up an additional pcie3 / lga1155 mobo and drop the 3570k/gtx660 in as a gaming/htpc. I might be selling the OE i3 box soon, and if so I'm really hoping the new pivos is released by then
post #2163 of 3346
Thread Starter 
Yeah my media server has an Asrock Extreme3 Z77 in it now. I can drop in any IVY bridge (3570k for $149 seemed like good deal)
post #2164 of 3346
In regards to the above, I'll have to search down an additional cpu cooler that works well for the 3770k with some overclocking headroom. What are any of you out there using (looking med-high end for this)??
post #2165 of 3346
Thread Starter 
Water or Air ? For Air I just use the EVO212 ($30) It takes a little room but cools well. And you can add a second fan on it (120mm) if you want. That should get you 4.2ghz. That is the poor mans simple solution and the improvements with an $80+ cooler just aren't worth the price increase. It's very minimal difference at all.

I hit 4.8ghz on mine before it was getting too hot (4770k) Prior to this I had a 2600k. I skipped the 3770k so no first hand experience with that.

Otherwise if you have big budget you can go water cooling which should get you a bit higher. You won't really be able to go more than 4.4ghz with your CPU unless you want to De-lid it. If you are going to get that serious you might as well go water cooling as well.

There was a good thread around here when I first starting clocking my i7 with videos and results from deliding, and some benches. I'll dig it up if you want. There is a couple dudes around here rocking 3770k and overclocking them. I do it more for sport than need. My 4770k is a beast.
post #2166 of 3346
what board are you running your 4770k on? i might need some help tweaking/overclocking. its running stock right now.
post #2167 of 3346
Thread Starter 
Quote:
Originally Posted by aliaskary77 View Post

what board are you running your 4770k on? i might need some help tweaking/overclocking. its running stock right now.

My 4770K is on Asrock Extreme 4. My i7 2600k Sandy Bridge was on an Asus Z68 Deluxe. I have an Asrock Extreme 3 Z77 as my server motherboard; I have a 3570K Ivy Bridge on PRO 3 Z77 Asrock for my HTPC (I think). They are all basically the same. Once you get the general idea anyways.

Overclocking on the Z87 is a bit trickier if you want to get extreme because Z87 and Haswell uses adaptive voltage offsets. Basically it can increase or decrease automatically the voltage in offsets related to different factors. This means if the offset calls for more voltage and you also apply more voltage you can give too much (too hot) or too little (instability). With Adaptive Mode, you don't set the specific voltage that you want in Windows, Adaptive allows the adoption on the top of what the Intel SVID (i.e.The Intel voltage rule to compensate the difference in CPU Ratio and CPU Turbo Ratio.) defines during production. The point of it is that it allows you to run a low voltage under idle and then under load the voltage increases and maintains stability. The benefit is power savings at idle without dropping CPU clocks like EIST would.

Manually doing it is the easy way. But you must set it for a single voltage (constant) based on your max clock - and this will be in effect at all times (wastes energy, causes heat) The stock settings allow for a down clock in not just clock speed, but also in voltage. That saves lots of energy and makes the chip efficient. Overclocking basically ruins all that and makes your CPU a total pig with energy. That's just how it works though. Adaptive basically allows you to take the "normal" voltage and apply more on top of that based on what is needed (or wanted) to allow for higher clocks (overclocking)

There was a couple of good videos on Newegg (ASUS) that explain overclocking on Z87 pretty well. AJ from Asus basically walks you through the basics of overclocking in that video and also introduces adaptive voltage.

Many motherboards come with the "auto tune" software that lets you overclock inside the software and basically sets the settings for you. That's not a bad option for mild overclocks, but it can get kinda crazy if you want to go extreme and it's typically less stable. The offsets are actually controlled inside the software on Z87 now though... so by relation the software overclock option is more viable on Z87 than it ever was on Z68 or Z77. This is because many people don't understand the voltage stair stepping offsets on Z87 and treat it like the older chips, and the software actually does to a mild extent understand it. Historically the auto tune software was a really terrible way to overclock versus doing it manually in the bios but I actually think on Z87 it's not a bad option for a noob. It also teaches you the basics so you understand what is going on in the bios better too.

4770k (and 3770k) run pretty darn hot. Like hot -- I mean really stupid hot. It's a terrible design from a heat dissipation perspective. If you want to get serious about overclocking on the new Intel chips you need to DE LID your CPU. Basically take a hammer to it (yes I am serious) and remove the cover (the top lid) and then put it back together with a superior thermal paste.





Some people can just get totally stupid with overclocking:




Dat Dere multiplier : Raise it. ^ biggrin.gif

Overclocking is fun because you are "stealing" more performance than you paid for. That's always a good time. I've done some pretty stupid things with CPU's over the years. I've bounced them off thermal throttling temperature limits many times... and from what I can tell most CPU's seem to enjoy it. Seriously, I have blatently abused most of my CPU's I've ever owned with reckless abandon and I've never had one fail yet. I'm not sure it's even possible on a modern CPU because they have protection built in to prevent them from melting in the hands of idiots. People always make it seems like overclocking is a big deal. It's really not. You just crank up the multiplier (or bus speed) to get a higher clock speed. Often you crank up the voltage to make it stable. Cranking the voltage does produce more heat and use more energy, but hey-- that's the point ! Just keep it under 100 degrees (Celsius, not Fahrenheit so yeah that's pretty friggin hot) and you should be fine. If you start going over 85c you should look at improving your cooling before you go further (higher clocks / voltage )

Setting the "TURBO MODE" to hit say 4.4ghz instead of the stock setting on something like a 4770k is an easy way to get extra performance when you need it, but not all the time like a manual overclock. This allows the CPU to drop down when it's not needed (clock speed and voltage drop) and save energy and run cooler. This is usually a nice compromise versus a standard manual overclock.
post #2168 of 3346
thanks. yeah things have changed a lot from the last time i had something to overclock...pentium 2 days i think.

i have the 4770k and the asrock z87 extreme 4. im on air with noctua NH-U12P. wouldnt go crazy but 4.0 or 4.2 would be nice. auto does seem like the best bet. never played with the a-tuning software thats comes with the board. defintely will be going the noob route.

update: it seems like it settled on 4.3, but will have to check for settings for that to be max, and not 100% of the time.
Edited by aliaskary77 - 12/4/13 at 8:27pm
post #2169 of 3346
Quote:
Originally Posted by Mfusick View Post

My 4770K is on Asrock Extreme 4. My i7 2600k Sandy Bridge was on an Asus Z68 Deluxe. I have an Asrock Extreme 3 Z77 as my server motherboard; I have a 3570K Ivy Bridge on PRO 3 Z77 Asrock for my HTPC (I think). They are all basically the same. Once you get the general idea anyways.

Overclocking on the Z87 is a bit trickier if you want to get extreme because Z87 and Haswell uses adaptive voltage offsets. Basically it can increase or decrease automatically the voltage in offsets related to different factors. This means if the offset calls for more voltage and you also apply more voltage you can give too much (too hot) or too little (instability). With Adaptive Mode, you don't set the specific voltage that you want in Windows, Adaptive allows the adoption on the top of what the Intel SVID (i.e.The Intel voltage rule to compensate the difference in CPU Ratio and CPU Turbo Ratio.) defines during production. The point of it is that it allows you to run a low voltage under idle and then under load the voltage increases and maintains stability. The benefit is power savings at idle without dropping CPU clocks like EIST would.

Manually doing it is the easy way. But you must set it for a single voltage (constant) based on your max clock - and this will be in effect at all times (wastes energy, causes heat) The stock settings allow for a down clock in not just clock speed, but also in voltage. That saves lots of energy and makes the chip efficient. Overclocking basically ruins all that and makes your CPU a total pig with energy. That's just how it works though. Adaptive basically allows you to take the "normal" voltage and apply more on top of that based on what is needed (or wanted) to allow for higher clocks (overclocking)

There was a couple of good videos on Newegg (ASUS) that explain overclocking on Z87 pretty well. AJ from Asus basically walks you through the basics of overclocking in that video and also introduces adaptive voltage.

Many motherboards come with the "auto tune" software that lets you overclock inside the software and basically sets the settings for you. That's not a bad option for mild overclocks, but it can get kinda crazy if you want to go extreme and it's typically less stable. The offsets are actually controlled inside the software on Z87 now though... so by relation the software overclock option is more viable on Z87 than it ever was on Z68 or Z77. This is because many people don't understand the voltage stair stepping offsets on Z87 and treat it like the older chips, and the software actually does to a mild extent understand it. Historically the auto tune software was a really terrible way to overclock versus doing it manually in the bios but I actually think on Z87 it's not a bad option for a noob. It also teaches you the basics so you understand what is going on in the bios better too.

4770k (and 3770k) run pretty darn hot. Like hot -- I mean really stupid hot. It's a terrible design from a heat dissipation perspective. If you want to get serious about overclocking on the new Intel chips you need to DE LID your CPU. Basically take a hammer to it (yes I am serious) and remove the cover (the top lid) and then put it back together with a superior thermal paste.





Some people can just get totally stupid with overclocking:




Dat Dere multiplier : Raise it. ^ biggrin.gif

Overclocking is fun because you are "stealing" more performance than you paid for. That's always a good time. I've done some pretty stupid things with CPU's over the years. I've bounced them off thermal throttling temperature limits many times... and from what I can tell most CPU's seem to enjoy it. Seriously, I have blatently abused most of my CPU's I've ever owned with reckless abandon and I've never had one fail yet. I'm not sure it's even possible on a modern CPU because they have protection built in to prevent them from melting in the hands of idiots. People always make it seems like overclocking is a big deal. It's really not. You just crank up the multiplier (or bus speed) to get a higher clock speed. Often you crank up the voltage to make it stable. Cranking the voltage does produce more heat and use more energy, but hey-- that's the point ! Just keep it under 100 degrees (Celsius, not Fahrenheit so yeah that's pretty friggin hot) and you should be fine. If you start going over 85c you should look at improving your cooling before you go further (higher clocks / voltage )

Setting the "TURBO MODE" to hit say 4.4ghz instead of the stock setting on something like a 4770k is an easy way to get extra performance when you need it, but not all the time like a manual overclock. This allows the CPU to drop down when it's not needed (clock speed and voltage drop) and save energy and run cooler. This is usually a nice compromise versus a standard manual overclock.
Yep I agree with this. People used to, and still do I suppose, post up how chip life would be vastly lessened when overclocked. Well, I'm still running a q6600 that I oc'd when it was new. It has practically ran at 100% load across all cores for 24/7 since then also. I reboot it every now and then just cause it's windows haha. It's still going strong.
post #2170 of 3346
Thread Starter 
Quote:
Originally Posted by DotJun View Post

Yep I agree with this. People used to, and still do I suppose, post up how chip life would be vastly lessened when overclocked. Well, I'm still running a q6600 that I oc'd when it was new. It has practically ran at 100% load across all cores for 24/7 since then also. I reboot it every now and then just cause it's windows haha. It's still going strong.

All my CPU always lasted me longer than I wanted to own them. CPU get replaced often with something newer and better. 5 years ago I had an E8500 Core2 DUO I paid $315 for it. It was DDR2 800mhz on Socket 775. Today that $300 Asus Deluxe motherboard and $315 CPU, with 4 sticks of Crucial Ballistix OC DDR2 is worth almost nothing. I would take a $35-$50 CPU and $50 motherboard over it without question, as well as 4GB in two sticks of 2GB DDR3 versus 4 sticks of 1GB each DDR2.

I ran that CPU on air @3800mhz for a long time. It lasted.

I just don't think CPU break or fail very often, I usually replace them because I WANT A NEW ONE long before they die. I don't expect much more than 3 years from my PC (or CPU) as three generations is about as long as they stay viable. If you go out and spend a few hundred dollars or more on a CPU you do expect to keep it for years- but eventually the point will come when a $40 CPU is better. At that point, I guess it just doesn't matter much.

I never had a CPU fail on me. Never. Not even once. Never RMA-ed a CPU. Seems like something else always breaks or fails before it.

Quote:
Originally Posted by aliaskary77 View Post

thanks. yeah things have changed a lot from the last time i had something to overclock...pentium 2 days i think.

i have the 4770k and the asrock z87 extreme 4. im on air with noctua NH-U12P. wouldnt go crazy but 4.0 or 4.2 would be nice. auto does seem like the best bet. never played with the a-tuning software thats comes with the board. defintely will be going the noob route.

update: it seems like it settled on 4.3, but will have to check for settings for that to be max, and not 100% of the time.

4.3 is good for you on air I think. You are safe there. Multiplier 43, bump the voltage offset (or run it like 1.22v manually)

1.26v should get you 4500mhz.

You could give it 2.56v if you are feeling really lucky eek.gif

post #2171 of 3346
Quote:
Originally Posted by Mfusick View Post

All my CPU always lasted me longer than I wanted to own them. CPU get replaced often with something newer and better. 5 years ago I had an E8500 Core2 DUO I paid $315 for it. It was DDR2 800mhz on Socket 775. Today that $300 Asus Deluxe motherboard and $315 CPU, with 4 sticks of Crucial Ballistix OC DDR2 is worth almost nothing. I would take a $35-$50 CPU and $50 motherboard over it without question, as well as 4GB in two sticks of 2GB DDR3 versus 4 sticks of 1GB each DDR2.

I ran that CPU on air @3800mhz for a long time. It lasted.

I just don't think CPU break or fail very often, I usually replace them because I WANT A NEW ONE long before they die. I don't expect much more than 3 years from my PC (or CPU) as three generations is about as long as they stay viable. If you go out and spend a few hundred dollars or more on a CPU you do expect to keep it for years- but eventually the point will come when a $40 CPU is better. At that point, I guess it just doesn't matter much.

I never had a CPU fail on me. Never. Not even once. Never RMA-ed a CPU. Seems like something else always breaks or fails before it.
4.3 is good for you on air I think. You are safe there. Multiplier 43, bump the voltage offset (or run it like 1.22v manually)

1.26v should get you 4500mhz.

You could give it 2.56v if you are feeling really lucky eek.gif

Yep totally agree with you. It amazes me that people still make grand claims on CPU degradation due to being overclocked. I've proven multiple times that is not the case, though I suppose since it's just me the control group is small haha
post #2172 of 3346
Thread Starter 
Quote:
Originally Posted by DotJun View Post

Yep totally agree with you. It amazes me that people still make grand claims on CPU degradation due to being overclocked. I've proven multiple times that is not the case, though I suppose since it's just me the control group is small haha

I don't think it's just you. I think anyone that has overclocked CPU's before realizes it. Usually it's the folks that have never done it, or don't even own unlocked chips that cry about such things.

The CPU becomes unstable long before you damage it. Who is going to overclock a CPU and let it BSOD constantly ? That makes no sense. If you are intelligent about it, then it's safe. Heat might kill a CPU but it's just one of those things it might not too. There is no moving parts so your only worry is burning it out. That's just unlikely from my experience, even if you abuse it. CPU can take the abuse.
post #2173 of 3346
Degradation depends on how much vcore you give the chip not how high the frequency is. Every chip has a two walls. The first is the wall it hits with normal vcore and the second is the wall it hits no matter what vcore. Pretty much everyone who has run chips between those two walls over their overclocking years tends to agree that their capable upper limit at a given vcore degrades over time. The chip may be stable for a year or more but at some point you hit a BSOD and have to lower the clock, vcore or both to get it stable again. Time passes and it happens again. You keep lowering the clock overtime and eventually it won't even run at stock settings. itznfb proved this by killing about 15 3770k's over the time span of about 8~9 months by running them on water at 1.6vcore. (I don't remember the exact number or time frame so don't quote me on it). The temps were kept at normal range (~70c) and ran 100% CPU 24/7 (WCG). He has a spreadsheet on a forum somewhere and every chip seemed to degrade at about the same rate. I think only 1 or 2 of them survived. Other people have done similar tests with the same results and the theoretical conclusion is that any chip running at any vcore is going to degrade over time. Running at it's rated vcore that may take 10 years to see the degradation who knows. I guess they would need to run more tests at varying vcore to see if the difference would be linear to judge how long it would take to degrade at stock vcore.

I'm not trying to argue any point or prove anything I just thought you might find that information interesting. As those settings are way out of scope of this conversation anyway. I wouldn't exactly call those kinds of tests definitive but they do provide a useful and informative data point.
post #2174 of 3346
Thread Starter 
Quote:
Originally Posted by techmattr View Post

Degradation depends on how much vcore you give the chip not how high the frequency is. Every chip has a two walls. The first is the wall it hits with normal vcore and the second is the wall it hits no matter what vcore. Pretty much everyone who has run chips between those two walls over their overclocking years tends to agree that their capable upper limit at a given vcore degrades over time. The chip may be stable for a year or more but at some point you hit a BSOD and have to lower the clock, vcore or both to get it stable again. Time passes and it happens again. You keep lowering the clock overtime and eventually it won't even run at stock settings. itznfb proved this by killing about 15 3770k's over the time span of about 8~9 months by running them on water at 1.6vcore. (I don't remember the exact number or time frame so don't quote me on it). The temps were kept at normal range (~70c) and ran 100% CPU 24/7 (WCG). He has a spreadsheet on a forum somewhere and every chip seemed to degrade at about the same rate. I think only 1 or 2 of them survived. Other people have done similar tests with the same results and the theoretical conclusion is that any chip running at any vcore is going to degrade over time. Running at it's rated vcore that may take 10 years to see the degradation who knows. I guess they would need to run more tests at varying vcore to see if the difference would be linear to judge how long it would take to degrade at stock vcore.

I'm not trying to argue any point or prove anything I just thought you might find that information interesting. As those settings are way out of scope of this conversation anyway. I wouldn't exactly call those kinds of tests definitive but they do provide a useful and informative data point.

I do agree with this ^ I guess everything I was suggesting should be prefaced with me explaining I'm only a mild overclocker, and I do it intelligently with voltages that are considered reasonable by most. I would't run more than 1.3v chasing higher clocks, I find there is a point of diminishing returns.

A small bump in voltage (safe bump) can get you from the 3000-3500mhz range to 4000-4300mhz range and keep you pretty safe. But to go higher and higher it becomes harder and harder (that's my experience anyways). Stealing the first 1000mhz (which is a serious speed boost) is easy, but stealing the last few hundred takes some fine tuning, even higher temps and voltages, and some luck of the die (good chip).

I bought my 4770k and seemed to run 4400mhz on first shot without any troubles. I bought it when the chip launched so I was pretty new. Over the next couple weeks I noticed online everyone complaining that the chip sucks, you can't overclock it, etc.. I was like "WTF?" Seemed fine to me. I guess from what I read I got a good CPU.

BTW- Where is itznfb ??? Not seen him around here in a month.
post #2175 of 3346
Quote:
Originally Posted by Mfusick View Post

I do agree with this ^ I guess everything I was suggesting should be prefaced with me explaining I'm only a mild overclocker, and I do it intelligently with voltages that are considered reasonable by most. I would't run more than 1.3v chasing higher clocks, I find there is a point of diminishing returns.

A small bump in voltage (safe bump) can get you from the 3000-3500mhz range to 4000-4300mhz range and keep you pretty safe. But to go higher and higher it becomes harder and harder (that's my experience anyways). Stealing the first 1000mhz (which is a serious speed boost) is easy, but stealing the last few hundred takes some fine tuning, even higher temps and voltages, and some luck of the die (good chip).

I bought my 4770k and seemed to run 4400mhz on first shot without any troubles. I bought it when the chip launched so I was pretty new. Over the next couple weeks I noticed online everyone complaining that the chip sucks, you can't overclock it, etc.. I was like "WTF?" Seemed fine to me. I guess from what I read I got a good CPU.

BTW- Where is itznfb ??? Not seen him around here in a month.

Yeah luckily since the Core days we've seen relatively easy to overclock chips that do very well at stock vcore. The extreme voltages aren't needed anymore unless your benchmarking competitively or just want a really fast PC. As you see though it varies by batch. Some chips just don't OC at all and some hit 4.6~4.8 on stock vcore.

itznfb was banned back in October. I'm not sure if it was permanent or not but he just decided not to come back. Assassin stopped posting on the same day and he was involved in the same thread so I'm not sure if he was banned as well or just left.
post #2176 of 3346
The big question is whether there is any use to overclocking such a small bump in speed. What gain is there to it? CPUs are already well overpowered for what people need, and most things seem to be GPU limited.
post #2177 of 3346
Quote:
Originally Posted by cybrsage View Post

The big question is whether there is any use to overclocking such a small bump in speed. What gain is there to it? CPUs are already well overpowered for what people need, and most things seem to be GPU limited.

I know most people don't re-encode their movies but it makes a huge difference in re-encoding times. Plus you used to be able to buy a $100 CPU and get $1000 CPU performance out of it. That isn't really possible today with limited bus overclocking and needing a K series CPU to get an unlocked multiplier.

So you're right in the sense that in order to get a CPU (Intel anyway) that is capable of overclocking via multiplier then that CPU can already do the things you need it to do without overclocking. Something like a i3-4130K at a $150 price point would be an amazing chip as long as you didn't need 4 physical cores. I think now that it's so easy to overclock if they still offered the ability with the lower end chips no one would buy the high end i5's and i7's.
post #2178 of 3346
I don't reencode (I use ISOs), but how much time can be saved by increasing the core speed by 500MHz, for example? I really have no idea.
post #2179 of 3346
Quote:
Originally Posted by cybrsage View Post

I don't reencode (I use ISOs), but how much time can be saved by increasing the core speed by 500MHz, for example? I really have no idea.

I'm not sure. I don't think it's a linear gain either. I've only really done re-encoded at stock 3.9Ghz (3770k) and 5.1GHz. Going back to my previous post about degradation though... unfortunately my 3770k can no longer hit even 4.8GHz. I think the best I can get with it is 4.6GHz right now. It's now one of my servers running at stock settings.

I can say the difference between 3.9GHz and 5.1GHz was typically 6-10 hours savings. Pretty significant when some encodes run over 30 hours at stock speeds.
post #2180 of 3346
That is a huge time savings there! That would be worth it if you encode a lot.
post #2181 of 3346
If you encode a lot and the time bothers you, get a extreme edition. biggrin.gif
post #2182 of 3346
Quote:
Originally Posted by Mfusick View Post

I don't think it's just you. I think anyone that has overclocked CPU's before realizes it. Usually it's the folks that have never done it, or don't even own unlocked chips that cry about such things.

The CPU becomes unstable long before you damage it. Who is going to overclock a CPU and let it BSOD constantly ? That makes no sense. If you are intelligent about it, then it's safe. Heat might kill a CPU but it's just one of those things it might not too. There is no moving parts so your only worry is burning it out. That's just unlikely from my experience, even if you abuse it. CPU can take the abuse.
Totally agree. Even if my chip degrades faster, I'm going to replace it with something newer years before the chip dies. Also, I don't just oc and stop when it is stable. It has to be stable AND correct. No point in having a fast stable oc if your chip outputs 2 + 2 = 5
post #2183 of 3346
Quote:
Originally Posted by techmattr View Post

Degradation depends on how much vcore you give the chip not how high the frequency is. Every chip has a two walls. The first is the wall it hits with normal vcore and the second is the wall it hits no matter what vcore. Pretty much everyone who has run chips between those two walls over their overclocking years tends to agree that their capable upper limit at a given vcore degrades over time. The chip may be stable for a year or more but at some point you hit a BSOD and have to lower the clock, vcore or both to get it stable again. Time passes and it happens again. You keep lowering the clock overtime and eventually it won't even run at stock settings. itznfb proved this by killing about 15 3770k's over the time span of about 8~9 months by running them on water at 1.6vcore. (I don't remember the exact number or time frame so don't quote me on it). The temps were kept at normal range (~70c) and ran 100% CPU 24/7 (WCG). He has a spreadsheet on a forum somewhere and every chip seemed to degrade at about the same rate. I think only 1 or 2 of them survived. Other people have done similar tests with the same results and the theoretical conclusion is that any chip running at any vcore is going to degrade over time. Running at it's rated vcore that may take 10 years to see the degradation who knows. I guess they would need to run more tests at varying vcore to see if the difference would be linear to judge how long it would take to degrade at stock vcore.

I'm not trying to argue any point or prove anything I just thought you might find that information interesting. As those settings are way out of scope of this conversation anyway. I wouldn't exactly call those kinds of tests definitive but they do provide a useful and informative data point.
I understand this, but I just assumed anyone doing any kind of oc'ing knows that clock increase and voltage increase go hand in hand. I mean, I'm not saying my chip has been running for years while I dump 1.21 gigawatts into it 😊
post #2184 of 3346
Thread Starter 
I am just bumping this for a buddy so I can link him:


Guide to Flash IBM 1015 to IT mode [Converting an LSI9240(IBM M1015) to a LSI9211-IT mode]


First make a USB bootable stick so you can boot into DOS command prompt.
You can do so easily here: http://www.sevenforums.com/tutorials/46707-ms-dos-bootable-flash-drive-create.html


Once you have that you need to copy the files onto the USB stick so you can reflash your IBM card with them.
First download from here:

http://forums.laptopvideo2go.com/topic/29059-sas2008-lsi92409211-firmware-files/

first do this:



Inside the folder looks like this:



then inside that looks like this:



This ^ above is what you want on your USB stick. It is really all you should see since the OS files for DOS are hidden and do not show.

Hardware matters: (make sure you have x8 slot and speed)
Find an older motherboard without modern graphical BIOS for best results, and make sure the board your using can run x8 in the slot the card is installed into. If you have a choice install the IBM 1015 into the top or first slot for best results. Many motherboards have multiple x8 or x16 slots (I used x16 just like a video card would) since x8 cards fit perfectly into x16 video card slots with simply extra room unused in the x16 slot. Caution: Many motherboards with video cards will not run x8 speed when a x16 video card is also installed. You’ll need a fairly high end motherboard to have such features if your boards are older. Otherwise a motherboard without a video card is preferable IMO. I used an Asus LGA775 ROG Deluxe with a Core2duo E8500 in it. Being a ROG board it could run x8 speed on the second slot and I flashed the card successfully from beginning to end using this board after failing on several others.


The next step is to boot up into DOS (command prompt)

Once you have booted successfully to DOS using your bootable USB stick and are seeing C:\
Type: megarec -writesbr 0 sbrempty.bin [hit enter]
[The system will do a bunch of things and come back to C:\]





Type: megarec -cleanflash 0 [hit enter]
[The system will do a bunch of things and come back to C:\]
REBOOT MANUALLY and boot back to USB stick







Type: sas2flsh -o -f 2118it.bin -b mptsas2.rom (sas2flsh -o -f 2118it.bin if OptionROM is not needed) [hit enter]
[The system will do a bunch of things and come back to C:\]
*note it is a o and not a 0 (it is not a zero)
** You probably do not want an OPTION ROM unless your going to boot from it. It is safer and smarter to not use OPTION ROM and to use your OS drive from a motherboard SATA port IMO. Having no option ROM also allows you to just quickly boot up when restarting and there is not screen in boot up for your IBM card. I like it this way.






Type: sas2flsh -o -sasadd 500605bxxxxxxxxx (x= numbers for SAS address) [hit enter]
[The system will do a bunch of things and come back to C:\]
*There is a green (almost white) Sticker on your card with this address.
**it is a o and not an 0 (not a zero but a letter)
***There should be no spaces and no dashes when entering in the sasadd number even though they are on the sticker and card.



REBOOT MANUALLY and boot back to USB stick
Now you're Done!







I have expanded this:
Quote:
Convert LSI9240(IBM M1015) to a LSI9211-IT mode
Type in the following exactly:
megarec -writesbr 0 sbrempty.bin
megarec -cleanflash 0

sas2flsh -o -f 2118it.bin -b mptsas2.rom (sas2flsh -o -f 2118it.bin if OptionROM is not needed)
sas2flsh -o -sasadd 500605bxxxxxxxxx (x= numbers for SAS address)

Done!

I thought this^ above was a little less clear so I provided the extra in the steps above for basic help so a newbie will know what to do without making the mistakes that are common.

That is basically all it takes. It takes about 10 minutes total assuming you have compatible hardware. If you do not have compatible hardware your will see errors or not the screens above with "success" after each step.


If you see these screens or anything other than the ones I put above each step it probably means your motherboard your using is not a good choice. I'd suggest finding another one and trying again from the first step. I messed up a bunch of times everytime I have flashed one of these and it always turned out great in the end. I'm not sure you can even brick a card. I think you could wipe it clean and flash it even if you did. But YMMV so be careful and follow the steps exactly.





good luck!
post #2185 of 3346
Thread Starter 
Quote:
Originally Posted by Nevcairiel View Post

If you encode a lot and the time bothers you, get a extreme edition. biggrin.gif

Or a dual Intel Xeon E5-2687W v2 machine, which should slap the sh!t out of an i7 tongue.gif

32 cores (16 logical and 16 Virtual) all running 5000mhz... Damn that would be sweet biggrin.gif

http://www.cpubenchmark.net/cpu.php?cpu=Intel+Xeon+E5-2687W+v2+%40+3.40GHz&id=2045
post #2186 of 3346
Thread Starter 
Quote:
Originally Posted by jbcain View Post

10-4.


if you were me, would you set the 4tb as parity (duh) and then set up the 4 new 3tbs into the pool and then put the 3 full 3tbs into the server, but not into the pool, and copy all the data into the existing drive pool, then wipe the drives clean, and then add them to the pool? (super dooper run on sentences suck). or would you just add the 3 full dries to the pool as is (i'm scared to lose data)

You won't lose the data. biggrin.gif

But I probably would copy then add empty into pool after I wiped them. Because I usually copy into specific locations and want specific organization. If adding the folder exactly has it is won't screw up the top level organization of your pool (can you tell I am a bit anal here) then sure go ahead. I find it easier to add empty drives and not suffer the parity calculation time (even though copy paste takes nearly as long first) - adding empty drives is nearly instant. I just feel more in control, but I am weird. I have added full drives many times to my pool. It's easy and you will not lose any data. Do whatever is easier.

No reason you can't add to the pool if you want to do it.
Quote:
Originally Posted by jbcain View Post

So, my 3 3tbs each have folders on the first layer, blurays, dvds, music, then subsequent folders in those. If I make the flex pool with the other drives then copy these drives to that pool, how does flex decide what info goes on what drive?

Basically everything that is on all your hard drives in your pool shows up on your pool. That is why it can get messy. For this reason I like to keep the top level of my pool organized.

I have folders like: HD MOVIES, TV SHOWS, DVD RIPS, BR RIPS, DISNEY, CHRISTMAS, MUSIC, ... etc...

Inside each of those folder would be that media. Example: All my TV shows are inside the "TV shows" folder. With flexraid it's possible (and likely) that some of your media collections will span multiple HDD's, especially if your collection is larger than any of your hard drives. This means you might have a "TV SHOWS" folder on several of your hard drives. The folder name is exactly the same on all of your hard drives that contain it. If you unpool your drives you would see a "TV SHOWS" folder on many different hard drives- but when you pool them you only see one folder on your FLEXRAID drive.

So answering your question - if you have multiple folders on those hard drives you add and the folders are named the same (or any hard drives contain folders named the same on the top level) then the contents would be pooled. If they are named different then all the folders would show on your pool and nothing would be pooled.

Flexraid doesn't move your data so it won't decide anything. If it's on a hard drive then it stays on that hard drive. The only time flexraid decided where stuff goes is when you are copying new data from outside your pool into your pool. In this case there is two settings in the set up- most times people like consolidated. Meaning Flexraid tries to keep everything together. If you already had a folder named "TV SHOWS" and you added more shows into that folder flexraid would put the new shows inside the folder on the drives that already contain a folder called "TV SHOWS" starting with the one with the most space first. If they all fill up flexraid then makes another "TV SHOWS" folder on a HDD that has the most available space and keeps going. It basically tries to limit the number of same folders you have by keeping it all together.

If you have drive balancing selected as an option it would not behave this way. It would try to balance how much data is stored on all your drives and fill them up together. This means anything new added always gets added to the drive with the most available space.

Lets use this example:


You have 4 hard drives in your pool. Only two of them have a folder called "MOVIES" The first hard drive with the folder "movies" is nearly full and only has about 100GB of free space out of 3TB. The second hard drive that has a movies folder is nearly full but has about 300GB of free space. The other two hard drives in your pool are both empty and contain nothing as you just added them to your pool. When pooled - You only see one "MOVIES" folder and the contents of both hard drives all show inside that folder (all your movies). Now lets say you copy and paste into your movies folder new movies and the copy paste is 500GB large. Flexraid would first start adding movies to the second drive which has 300GB of usable space- then if it fills it would begin to fill up the other movies folder until that drive was full (100GB more) and then if you still had more and both drives were filled it would make a new folder named exactly the same on the hard drive with the most available free space and begin filling that up.

Hope that makes sense.
post #2187 of 3346
Thread Starter 
Quote:
Originally Posted by spectrumbx View Post

1. http://wiki.flexraid.com/2011/04/15/quick-validate-vs-validate/

2. Scheduled entries cannot be edited. It is just much simpler to delete them and create new ones.

3. There is no documented typical scheduling because it all depends on your usage pattern.

Quick-Validate: Reports changes in your array. You can manually run it once here and there to check for changes, but no point in scheduling it.
I mean, changes in your array are expected.

Update: you run it as often as you need to (based on how often data is being changed in your array). You could run it every hour or once a day or once week. It is up to you to decide.
I also recommend running the Update task right after big data changes rather than waiting for the scheduled operation to trigger.

Validate: I run it once a week to detect datarot. You could run it once a month or any other frequency. It all depends on your level of paranoia. Make sure it runs only after an Update operation has run.

Verify: I run it once a month. Again, the frequency here is driven by your level of paranoia. Make sure it runs only after an Update operation has run.

Validate and Verify do the same thing but take a different approach in doing it.
Validate was designed to replace Verify as it runs faster and detects datarot whereas Verify just does bit for bit verification.

If you want more power, check out the FlexRAID Expression Language: http://wiki.flexraid.com/2011/06/02/the-flexraid-expression-language/

This topic is reminding me to add a task that automatically syncs the RAID first before executing Validate/Verify.
Made notes to Mantis: http://bug.flexraid.com/view.php?id=299



Windows 7 Home Premium 32Bit:
http://msft.digitalrivercontent.net/win/X17-58996.iso
Windows 7 Home Premium 64Bit:
http://msft.digitalrivercontent.net/win/X17-58997.iso
Windows 7 Professional 32Bit:
http://msft.digitalrivercontent.net/win/X17-59183.iso
Windows 7 Professional 64Bit:
http://msft.digitalrivercontent.net/win/X17-59186.iso
Windows 7 Ultimate 32Bit:
http://msft.digitalrivercontent.net/win/X17-59463.iso
Windows 7 Ultimate 64Bit:
http://msft.digitalrivercontent.net/win/X17-59465.iso

If your trying to install Windows home server to SSD smaller than 160GB here is how I do it:

Hit SHIFT +F10 to bring up command prompt. Then type : notepad.exe and hit enter. This will bring up the Notepad. Click file then open and browse to SKU/SERVERHOMEPREMIUM.def. You need to select all files at the bottom, and not just word files to see it. Then just edit the size of the HDD from the 160GB to anything you want it to be. Save and Close.

Then just type wpshell.exe and enter... set up will continue on your drive under 160GB smile.gif

post #2188 of 3346
Thread Starter 
I'm installing (re-installing) WHS 2011 to my 120GB SSD and I am ashamed to admit I forgot how and had to go look for that ^

Now I remember biggrin.gif
post #2189 of 3346
What's the benefit to flashing the m1015 card just for flexraid pooling? I thought it was just to gain raid 6
post #2190 of 3346
Thread Starter 
Quote:
Originally Posted by jbcain View Post

What's the benefit to flashing the m1015 card just for flexraid pooling? I thought it was just to gain raid 6

It is to remove RAID and boot screens and make it a transparent HBA card (basically just sata ports)

You can buy a card like that but it's more expensive. So flashing a cheaper card saves you money.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Home Theater Computers
AVS › AVS Forum › Video Components › Home Theater Computers › Mfusick's How to build an affordable 30TB Flexraid media server: Information Requested.!