More on CPU specs of the ReadyNAS Pro

Ok, I should really look a little more before asking questions..

I think the chipset on the flame6 mobo is a q963. There is certainly on board VGA and sound and lspci seems to suggest it could be..

00:00.0 Host bridge: Intel Corporation Q963/Q965 Memory Controller Hub (rev 02)

Now, checking intel's website for compatible chips I found:

http://ark.intel.com/Product.aspx?id=27 ... L9U2,SL9ZF
Intel® Core™2 Duo Processor E6700 (4M Cache, 2.66 GHz, 1066 MHz FSB)

Now, the current CPU is:
http://ark.intel.com/Product.aspx?id=29 ... LA3H,SLA8Z
Intel® Pentium® Processor E2160 (1M Cache, 1.80 GHz, 800 MHz FSB)

Doing a comparison between the two, on intel's site:

Don't think embedded matters, see below
E2160 is embedded
E6700 isn't

Hmmm.. speeeeed bump!!
E2160 1.8GHz 1MB-L2 800MHz
E6700 2.66GHz 4MB-L2 1066MHz

Might be a problem, but may be not, see below
E2160 Core voltage: 0.85 - 1.5v
E6700 Core voltage: 0.85 - 1.3625v

Good.. virtualbox / vmware fun.. might need a kernel recompile though
E2160 No VT-x
E6700 VT-x enabled

This is the make up of the chip, apparently.. LPGA is plastic or some such..they should be interchangable
E2160 PGA-775
E6700 LPGA-775

Key thing is that the E6700 is also rated at 65W, like the E2160.. so the heatsink should have no problems, right?

Only problem I can see is the top voltage is slightly lower.. If the mobo decides to run it at 1.5v, it'll likely cook it.. Not sure if the flame mobo will behave and run it at its correct voltage.. would be nice though..

I saw somewhere that someone was using an E6300.. The specs between the E6700 and E6300 aren't actually that different. The E6300 isn't embedded so presumably that isn't a problem on this mobo and it also appears to have the same top voltage of 1.3625v

Sooooo... the E6700 be a contender?! If the chipset is Q963 or possibly Q965 it won't support any 1333MHz chips.. pitty :( However getting a nice core2 in there would significantly help my transcoding exploits!!

It may work. We are not going to endorse any CPU mods, and of course, your warranty is at risk by doing that modification.

I am the one running an E6300 in my personal system at home. It has worked for me, but as always, YMMV (your mileage may vary). Don't need a kernel re-compile. I have VirtualBox running on mine.

I absolutely understand your stand point, the the urge to fiddle and play is too strong! :)

Not needing a re-compile is serious bonus! I don't fancy the idea of messing with the boot process. Unlike a PC, I can't stick in a CD/USB key and repair it if I break it :)

Is there any problem with the voltage ranges in the vpd information? It reports min and max voltages? If I stick in a lower voltage CPU, I'm guessing this won't matter. I suspected the values in the file are really only for monitoring and there doesn't appear to be a particular voltages min-max pair that would allow 1.5v (E2160) but reject 1.325v (E6xxx)

VT extensions would be great!

I saw the E7400 on sale ($79, newegg/amazon is $119), but then checked out the specs like you have linked above and noticed it has no VT-x ;(

Looking at the E7400 on intel's site, it doesn't list Q963 or Q965 as a supported chipset.. hence why I didn't try and find one to test..
I'm not totally certain which chipset is actually on the motherboard (FLAME6 I'm guessing it is very custom made for Netgear, google tells me nothing!) but I think it is one of those. Buying a chip that didn't list support for Q963/Q965 would be REALLY pushing my luck :)

I just received a E6600 2.4GHz chip in the mail.. I couldn't find an E6700 2.66GHz chip for sale without the risk of it being held up in customs for weeks so settled with the next one down at a fraction of the price (for the record, I suspect the E6700 *might* be the fastest the board will support)

I've whacked it in and it booted YEAY!! Will need to keep a check on the CPU temp.. the retail intel heatsink it shipped with is massive compared to the one in the pro so there could be some problems there.. I'll wait and see:-o

Either way, here's some stats: hdparm -t -T /dev/c/c

Stock Pro:
Timing cached reads: 1890 MB in 2.00 seconds = 945.13 MB/sec
Timing buffered disk reads: 510 MB in 3.00 seconds = 170.11 MB/sec

Pro with E6600 and 4GB of 800MHz RAM:
Timing cached reads: 9052 MB in 2.00 seconds = 4533.83 MB/sec
Timing buffered disk reads: 628 MB in 3.00 seconds = 208.99 MB/sec

WHAT THE?! The cached read is over 4 times faster!! FANTASTIC!!

I'm guessing the buffered read speed has kind of plateaued.

I'm going to leave it memtesting over night. It isn't a great test for CPU stability, but it is a start. I've not migrated anything over yet as I'm still developing a add-on interface for the stuff I need so I don't mind waiting to prove it is stable..

Damn, I wish I'd found an E6700, just think of the 2.6GHz goodness.. oh well.. Must be happy with what I have..

Nice, I might need to update mine too, though my VirtualBox VM runs at more than satisfactory rates for me at the moment, and I use it as a daily workhorse, my laptop at home is just acting as a thin client to it.

Ok. I happened to mention the upgrade to someone who was also after an upgrade.. Long and short, I ended up with an E6700 cpu. I've run the same hdparm -t -T /dev/c/c tests and there's a marginal improvement..

Stock Pro with E2160 (1.8GHz) and 1GB of 667MHz RAM:
Timing cached reads: 1890 MB in 2.00 seconds = 945.13 MB/sec
Timing buffered disk reads: 510 MB in 3.00 seconds = 170.11 MB/sec

Pro with E6600 (2.4GHz) and 4GB of 800MHz RAM:
Timing cached reads: 9052 MB in 2.00 seconds = 4533.83 MB/sec
Timing buffered disk reads: 628 MB in 3.00 seconds = 208.99 MB/sec

Pro with E6700 (2.66GHz) and 4GB of 800MHz RAM:
Timing cached reads: 10132 MB in 2.00 seconds = 5075.66 MB/sec
Timing buffered disk reads: 622 MB in 3.00 seconds = 207.27 MB/sec

Looks like IO is more of a problem and getting anything more than about 208MB/s out of the RAID6 array is going to be hard..

Transcoding is noticably faster though.. not that I have any real stats to go on yet..

Still, if I'm right and this is the fastest CPU you can put in a Pro, I am more than happy.. it is BLISTERINGLY fast!

If anyone wants to make me an offer for a E6600 2.4GHz chip, feel free :)

cpitchford wrote:
If anyone wants to make me an offer for a E6600 2.4GHz chip, feel free :)

$20 :)

how does the cooling keep up with it? i mean the cooler on the pro isnt exactly high end stuff and the faster processor produces way more heat for sure.

could you post temperatures and fan-speeds?

Yeah, it is funny you should ask!

I was seriously contemplating Chirpa's offer.. but

I have two tests..
for i in 1 2 3 4 5 6 7 8 ; md5sum /dev/zero & done
This just generates cpu load.. a specific type of cpu load, but load none the less

The other test is multiple compiles of the linux kernel ;)

Anyway, with the 8 copies of md5sum running I hit an overheat problem..

According to frontview -> system -> health, temp2 reached 61 degress C.. which is 1 over the CPU limit.. I tried patching the vpd file bumping it up to 65 (same as temp1) but it doesn't seem to get it's limits from the vpd info file in /proc....

Having a look in /sys for temp info I found that:

grep . `find /sys -name temp?_*`
/sys/devices/platform/coretemp.0/temp1_max:85000
/sys/devices/platform/coretemp.0/temp1_label:Core 0
/sys/devices/platform/coretemp.0/temp1_crit_alarm:0
/sys/devices/platform/coretemp.0/temp1_input:89000
/sys/devices/platform/coretemp.0/temp1_crit:100000
/sys/devices/platform/coretemp.1/temp1_max:85000
/sys/devices/platform/coretemp.1/temp1_label:Core 1
/sys/devices/platform/coretemp.1/temp1_crit_alarm:0
/sys/devices/platform/coretemp.1/temp1_input:89000
/sys/devices/platform/coretemp.1/temp1_crit:100000

Now, these values are the values reported by the processor itself (one per core).. The max value is 85 degress C.. but I'm still not sure what "max" means.. According to the kernel doc:

temp1_max - All cooling devices should be turned on (on Core2).

So, does that mean the CPU will be damaged if it is exceeded.. or does it just mean that the CPU fan must be running flat out if the CPU is this hot?

The reason I ask is that temp1_crit also exists.. and says 100 degrees C.

So.. with the E6700, the temp2 reading was 1 degree over the limit (I think this limit is imposed by Netgear) and the core's reported a temperature of 89 degrees C.. Hot hot hot..

I've whacked the E6600 back in:

/sys/devices/platform/coretemp.0/temp1_max:84000
/sys/devices/platform/coretemp.0/temp1_label:Core 0
/sys/devices/platform/coretemp.0/temp1_crit_alarm:0
/sys/devices/platform/coretemp.0/temp1_input:85000
/sys/devices/platform/coretemp.0/temp1_crit:100000
/sys/devices/platform/coretemp.1/temp1_max:84000
/sys/devices/platform/coretemp.1/temp1_label:Core 1
/sys/devices/platform/coretemp.1/temp1_crit_alarm:0
/sys/devices/platform/coretemp.1/temp1_input:86000
/sys/devices/platform/coretemp.1/temp1_crit:100000

Now, the temp1 is still higher than the "max" value.. I still don't know if that is a problem.. however. temp2 on frontview reports a MUCH healthier 53 degrees C

So.. the E6600 is 8 degrees cooler running flat out.. and is below the 60 degree limit.. both CPU cores report higher than max levels.. but both are lower than the "crit" level.. so I'm still working out (reading the source and CPU spec) what this means.

Slight pain is that the CPU cooler is attached with screws.. the backing plate is part of the case so heatsinks that screw on can be used.. Stock intel coolers, for example, with their push plugs cannot be used in this case.

If only the case was 1 CM wider a MUCH nicer fan could be put in..but from what I can tell, the tallest cooler you could use is 53mm and the selection is limited..

From what I can tell, the NAS is happy and relatively cool running an E6600, but an E6700 is going to be pushing it.. I've not tried a kernel compile reliability test or a memtest on the E6700 but I don't honestly think the CPU cooler is up to it..

If I can get a larger cooler, that uses screws rather than push plugs, I might try again.. but in the mean time 2.4Gig seems that safe limit with the stock readynas cooler.. (which is remarkably small compared to the intel stock ones! like less than half the height!)

Shame.. Still 2.4G is pretty damn good! :)

If I find out what the temp1_max actually means (ie can it be safely exceeded / temp1_crit is the REAL top limit) I'll let you all know