E7200 CPU - GA-P35-DS3

andyb · Post by **andyb** » Thu May 01, 2008 2:39 pm

I have an E7200, it is currently at my workplace running 2x prime 95's @ 3.16GHz overnight.

This overclock seems to be rather easy (as reviews have suggested), I have not needed to bump the voltage at all, it is still running at the default 1.1750v.

It runs very cool overclocked using my Xigmatek HDT-S1238 cooler in open air using the provided fan (950-1100 rpm). It runs at <42C in an ambient temp of ~20C with both cores flat out.

All being well, I plan on installing it im my PC tomorrow evening, does anyone want any specific benchmarks run to see how quick it is.

Seemingly the only difference between the E7200 overclocked and the E8xxx CPU's (non-overclocked) is the difference of 1MB of cache and a reasonable price saving.

Andy

XS Janus · Post by **XS Janus** » Thu May 01, 2008 3:17 pm

Will you do some undervolted power measurements?

andyb · Post by **andyb** » Thu May 01, 2008 3:24 pm

I can certaily do power measurements with the CPU clocked at default, and overclocked, whilst idle, 1x core @ 100% load and both cores @ 100% load.

I will consider underclocking at the CPU's default clock, but as I will have this running my already installed build of XP, and loads of Apps and games I dont want to jepordise the stability of XP.

Andy

fwki · Post by **fwki** » Thu May 01, 2008 4:05 pm

When you have it back to default settings, could you enable EIST and check idle VID or Vcore? And I am also curious what the idle multiplier drops to. CPU-z should show this data. At load the multiplier would be 9.5x so perhaps the p-states are 8.5, 7.5 etc.

angelkiller · Post by **angelkiller** » Thu May 01, 2008 4:12 pm

andyb wrote:does anyone want any specific benchmarks run to see how quick it is.

A 1M run of SuperPi (XS mod) would be great.
-Thanks

andyb · Post by **andyb** » Fri May 02, 2008 2:08 am

Here is an update + some answered questions.

As of this morning my e7200 was still running 2x Prime95 with no errors, Speedfan reported the 2 cores @ 37C with the fan running @ 985rpm.

When I stopped Prime95 the Cores dropped down to 28C, and the fan down to 810rpm.

CPU-Z tells me that the CPU has dropped down to x6 multiplier. As it is currently overclocked it runs @ 2,000 MHz, if it were running at its default 266MHz BUS it would be down to 1,600 MHz. The voltage dropped down to 1.168v from 1.152v

WTF (it actually went up according to CPU-Z). I just tested it again, and can confirm it wasnt a fluke.

Super Pi XS 1M 16.234 - 16.250 - 16.234 (3 times, although it says there is a bad checksum). I will check that out later. Previous versions managed to get the time wrong - it is difficult to have lots of faith in something that had such a large cockup, especially after 2 copies of Prime95 ran error free for 16.5 hours.

Andy

fwki · Post by **fwki** » Fri May 02, 2008 4:48 am

andyb wrote:The voltage dropped down to 1.168v from 1.152v WTF (it actually went up according to CPU-Z). I just tested it again, and can confirm it wasnt a fluke.
Andy

No it's not a fluke. Because you have manually set VID at 1.1750v, EIST is unable to change it. CPU-z is reporting the Vcore measurement after Vdroop and the drop in voltage from VID is higher at load. Accordingly, Vcore at idle is higher than that at load. I was curious what the voltages would be if cpu VID in BIOS was set to Auto and EIST could manage it, but you can't do that until clocks are back to default.

andyb · Post by **andyb** » Fri May 02, 2008 5:18 am

Because you have manually set VID at 1.1750v, EIST is unable to change it.

I did fiddle with the voltage a few times, but I am quite sure that the voltage is set to "Auto" in the BIOS (which on the E7200 is 1.175v).

When I get it installed, I will look at the voltage when at default CPU settings to see if there is any difference. I will also check that the voltage is on "Auto" and not manually set to 1.175v, and that EIST is set up correctly.

I understand that this CPU has very low idle power usage at default clock, and that the CPU can be undervolted as well - although as mentioned above, I dont really want to try this as I could end up corrupting my XP install.

Andy

jaganath · Post by **jaganath** » Fri May 02, 2008 12:07 pm

CPU can be undervolted as well - although as mentioned above, I dont really want to try this as I could end up corrupting my XP install.

fyi, i've never managed to corrupt an XP install thru undervolting in bios. usually all that's required is to reset CMOS if you go too low (although this can be fiddly on some boards).

ReelMonza · Post by **ReelMonza** » Fri May 02, 2008 1:09 pm

I already corrupted my XP install while undervolting.

jaganath · Post by **jaganath** » Fri May 02, 2008 3:34 pm

ReelMonza wrote:I already corrupted my XP install while undervolting.

software or bios?

ReelMonza · Post by **ReelMonza** » Fri May 02, 2008 6:04 pm

BIOS and software.
OS corrupted while booting

Mats · Post by **Mats** » Fri May 02, 2008 6:58 pm

andyb wrote:Seemingly the only difference between the E7200 overclocked and the E8xxx CPU's (non-overclocked) is the difference of 1MB of cache and a reasonable price saving.

The differences in stock settings are 3 MB cache instead of 6 MB, 266 MHz FSB instead of 333 MHz, and a PCB with a different layout that have a lower overclocking potential. Oh and price, of course.

andyb · Post by **andyb** » Tue May 06, 2008 2:40 pm

Here are some facts and figures for everyone.

Part of this is a testament to the cool running of the E7200 (even when overclocked by 25%), and part of this can be attributed to the excelent Xigmatek HDT-S1238 CPU cooler.

After some time of playing a real world game (BF2), with just Core 0 and both cores running the game the cores stay at 38C. This is with a Nexus @ 5v on the cooler, and the only case fan a Slipstream 1200 @ 950 rpm (damned annoying really).

I checked the BIOS, and I have to admit that I made a mistake, whilst going through the procedure of overclocking I had the voltage up to 1.200v as a precaution, and never dropped it down to "auto" afterwards.

The Core voltage at idle according to CPU-Z 1.44.2 is 1.120/1.136v (it changes every 2 seconds).

Whilst running Prim95 2x times CPU-Z reads the core voltage at 1.136v, I will have to check the BIOS again to see that everything is set correctly, or does the voltage not drop as dramatically as with AMD CPU's.?

The temp of the 2 cores are 45C and 43C, this is more than enough even in the height of summer, and I strongly suspect that it will be perfectly cool with the case fan @ 5v as well, which will be something for tomorrow. And if the Slipstream @ 5v displeases me I will hack my Nexus around so I can mount it using my silicone grommets.

Any more questions, or tests requested will be forthcoming.

The differences in stock settings are 3 MB cache instead of 6 MB, 266 MHz FSB instead of 333 MHz, and a PCB with a different layout that have a lower overclocking potential. Oh and price, of course.

OK, 3MB of cache, but when overclocked to 333 FSB the difference is not that great. One of the reasons why I selected the E7200 was that I could get a free performance boost with a cheapish chip and not have to worry about overclocking the RAM, chipset or motherboard. If I had bought an E8xxx CPU I would not have overclocked the CPU at all as all of the motherboards that support more than 333 FSB natively cost a mint.

Andy

juamez · Post by **juamez** » Wed May 07, 2008 6:23 am

andyb wrote: The Core voltage at idle according to CPU-Z 1.44.2 is 1.120/1.136v (it changes every 2 seconds).

Whilst running Prim95 2x times CPU-Z reads the core voltage at 1.136v, I will have to check the BIOS again to see that everything is set correctly, or does the voltage not drop as dramatically as with AMD CPU's.?

I have read that the Gigabyte P35 boards have very little to no drop in voltage while stressing the cpu. Which is nice, eh?

Voltage dropping (="Vdroop") is more dependant on the used motherboard/chipset than cpu brand in general, btw.

smilingcrow · Post by **smilingcrow** » Wed May 07, 2008 2:09 pm

juamez wrote:I have read that the Gigabyte P35 boards have very little to no drop in voltage while stressing the cpu.

My P35-DS3R Rev 2.0 has the following VDroop as reported by CPU-Z:

E4500
BIOS - 1.05V
Idle â€“ 1.024V
Load â€“ 1.008V

Q9300
BIOS - 1.037V
Idle â€“ 1.008V
Load â€“ 0.976V

amyhughes · Post by **amyhughes** » Wed May 07, 2008 5:52 pm

juamez wrote:I have read that the Gigabyte P35 boards have very little to no drop in voltage while stressing the cpu.

I've taken a lot of data at different voltages/frequencies with a GA-P35-DS3L and E8400 and I've found there is often no additional droop at load over idle.

Some examples...

300x6=1.8 GHz, 0.775 V set in BIOS, cpu-z reports 0.752 idle and load
333x6=2.0 GHz, 0.8125 V set, 0.784 reported idle and load
367x6=2.2 GHz, 0.85 V set, 0.816 reported idle and load
400x7=2.8 GHz, 0.975 V set, 0.944 reported idle and load
333x9=3.0 GHz, 1.0125 V set, 0.976 reported idle and load

Some counter-examples...

333x7=2.33 GHz, 0.875 V set, 0.848 reported idle, 0.832 load
400x6=2.4 GHz, 0.8875 V set, 0.864 reported idle, 0.848 load
400x9=3.6 GHz, 1.1625 V set, 1.136 reported idle, 1.120 load

andyb · Post by **andyb** » Mon May 12, 2008 3:55 pm

FYI have tested my CPU/mobo a little to check that I have some overhead.

As I dont plan on running my mobo/chipset/RAM out of spec, I dropped the RAM multiplier down to 2.5, and increased the BUS speed slightly (340, from 333, equaling a CPU clock of 2040, and RAM clock of 950MHz

This went through 8 hours of dual prime 95, and many hourse of BF2 and general usage with no ill effects over the last 3 days.

I am now satisfied that my system wont give me any overclocking related problems for some years

Now to clock down the BUS and up the RAM again.

Andy

wanky · Post by **wanky** » Thu May 15, 2008 11:05 pm

got one of these for a cheap upgrade, i like low voltage/low heat overlocking and this one made it stable up to
3325 @ 1.112v, temps 33 idle 52 load (tj 105) with Arctic Freezer 7

only thing pissing me off is asus qfan even in silent mode its ramping up fans to full speed under load causing so much noise..i wish it would only do that over 70 deg or something. 52 deg is still cool ffs

smilingcrow · Post by **smilingcrow** » Fri May 16, 2008 1:06 am

wanky wrote:only thing pissing me off is asus qfan even in silent mode its ramping up fans to full speed under load causing so much noise..i wish it would only do that over 70 deg or something. 52 deg is still cool ffs

Thatâ€™s why I donâ€™t trust BIOS based fan control and prefer to use Speedfan.

wanky · Post by **wanky** » Fri May 16, 2008 6:13 pm

smilingcrow wrote:
wanky wrote:only thing pissing me off is asus qfan even in silent mode its ramping up fans to full speed under load causing so much noise..i wish it would only do that over 70 deg or something. 52 deg is still cool ffs
Thatâ€™s why I donâ€™t trust BIOS based fan control and prefer to use Speedfan.

thx switched to speedfan works good with my mobo

mkk · Post by **mkk** » Sun May 18, 2008 11:52 am

I'll just chime in and note that with the latest beta BIOS(F13i) this GA-P35-DS3 motherboard seem to have regained an old cold boot(complete power off) voltage problem, which leads to the default voltage being used regardless of the manual setting. Now I have only tested with a higher than default voltage for overclocking so I can't say if the same occurrs when undervolting.

As my E7200 runs stably at 3.2GHz and default voltage I might just settle for that as it's so easily cooled, but I hope for a new BIOS for the future. If you experience the same, please take your time to mail Gigabyte about it.

ReelMonza · Post by **ReelMonza** » Sun May 18, 2008 1:30 pm

mkk, what is your "normal" CPU voltage in BIOS (the default CPU voltage) ?

mkk · Post by **mkk** » Sun May 18, 2008 2:56 pm

The BIOS claims the default is 1.2125V. The readout from the boards voltage meter in the BIOS reports a resulting voltage of 1.188. SpeedFan reads(presumably from the same sensor) an idle voltage of 1.12 and load about 1.152-1.68V... I'll reboot once and check those scores in the BIOS once more and edit here in a minute if I see something else.

To do 3.6GHz stably I had to use a setting of 1.35V, resulting reported to be about 1.31V last I checked.

bgavin · Post by **bgavin** » Mon May 19, 2008 7:24 am

I found the F13i broke my DS3 (not R) machine, so I backed down to F12 version.

The board came with the F6 bios, then I tried F13i, then back to F12 while trying to get rid of high DPC latency. None of these bios fixed the DPC problem. I do not have have the DS4 board, so I cannot use (I think) that bios that has the DPC temp fix installed.

One very annoying thing I noticed: the temps changed with every bios version. RealTemp 2.5, SpeedFan 4.34, HWMonitor 1.09 do not agree with each other. RT and SF are close, HWM is 10C higher, or more. As near as I can tell, RealTemp 2.5 seems to be the most accurate. I ran SpeedFan without calibration, so that may explain its discrepancy. All three got the hard disk temps correct.

The chronic BIOS problems with GA-P35 boards is a major weak spot in an otherwise exceptional board.

E7200 CPU - GA-P35-DS3

E7200 CPU - GA-P35-DS3

Re: E7200 CPU - GA-P35-DS3

Re: E7200 CPU - GA-P35-DS3