E2xxx and E4xxx overclock thread

[SerpentRoyal]

Originally posted by: 21stHermit

Originally posted by: Arkainium
so I removed the IHS. This helped a ton!

Help, what is a IHS?

Thanks
Hermit

Heat spreader.

 

[21stHermit]

Originally posted by: Arkainium
so I removed the IHS.

Clearly not for the faint of heart!!! :shocked:

 

[SerpentRoyal]

Not that hard if you have the right cutting tool. Took me about 45 minutes to do my E4300. Now I can pop-off the lid in about 20 min while watching TV. Not sure when I will have some time to mill the base of my BT so I can mount it directly to the CPU core. That should knock off another 5C from the load temp.

 

[brencat]

Originally posted by: Arkainium
Right now I'm stress testing at 3.6ghz (10x360) at 1.495V running orthos small ffts with temps peaking at 60C.

Excellent work Arkainium! Do a search for E2180 at www.overclock.net. Several really good threads to read. Guys are getting some really nice OCs with these chips.

I just ordered the E2180 myself an hour ago -- $83.70 at Buy.com with $10 Google checkout and free shipping. Woot! All I need now is the vid card to complete my build...

 

[Avalon]

Originally posted by: Arkainium
TAT doesn't seem to work with my setup; E2180 and Abit IP-35E. I get an error, "Error enumerating On Demand Clock Modulation support."

Anyway, initially I was using the stock heat sink which didn't do so bad actually. I managed to get to 3.2ghz (10x320) at 1.335V but my orthos load was peaking at 70C. At this point I already felt this chip was well worth the money, but I knew I could take it further.

So then I put on a TR Ultra-120 Extreme which brought the temps down a bit, but for the money I paid for it, I wasn't overly impressed. I was doing 3.33ghz (10x333) and orthos load was peaking at about 60C.

Then I decided to take it to the next level. So I lapped my hsf and cpu, but that really didn't help all that much. I got like 2-3C improvement.

Now this is where I just got foolish. I really wasn't expecting my processor to live through this, but I figured at $90 it wouldn't be a huge loss, so I removed the IHS. This helped a ton! When comparing stock to the Ultra-120 Extreme mounted directly on top of the core temps dropped by about 25-30C overall.

So now I'm trying to decide how high I want to take the vcore without frying my chip after a week. Right now I'm stress testing at 3.6ghz (10x360) at 1.495V running orthos small ffts with temps peaking at 60C.

I have a new power supply coming in tomorrow and I wonder if that'll change anything because the one I'm working with right now is a cheap 400w from my old computer. While OCing sometimes my computer restarts and I wonder if it's not enough vcore or my power supply can't keep up. I really only just got into overclocking.

If the new power supply doesn't improve my overclock, I'll probably stay at 10x360 if I deem it to be stable enough. I'll post an update after I settle and do some more testing.

Did you have to modify your TR Ultra Extreme to get it to seat properly on the bare core, or did it just work? I've got an E2140 and a TR Ultra-120, and am seeing some high temps...don't really care to lap, but if I can just hack off the IHS, I wouldn't mind giving that a shot if I didn't have to do any further modifications.

 

[Arkainium]

Originally posted by: Avalon
Did you have to modify your TR Ultra Extreme to get it to seat properly on the bare core, or did it just work? I've got an E2140 and a TR Ultra-120, and am seeing some high temps...don't really care to lap, but if I can just hack off the IHS, I wouldn't mind giving that a shot if I didn't have to do any further modifications.

The problem is the little swinging latch on the motherboard that locks the cpu to the socket sits too high to get good contact between the core and the heat sink. So I took that latch off but I still wasn't making good contact because the screws that come with the Ultra-120 Extreme don't go down all the way. So I used some washer-like things to make it tighter.

At first it wouldn't post, so I thought I messed up. But the problem is without the latch the cpu might not be secure so you need to be careful when you mount the heatsink without it. The latch comes off really easy, so it's just a matter of getting the heat sink on tight enough to hold the chip in place.

 

[lopri]

Why are folks seeing such high temps with these smaller L2 CPUs? (some even higher than quads?) I mean I have an E6600 (almost a year old now) that does 3.60GHz/1.29V/50C and I remember that my E6400 was even cooler than that with more voltages. E6600/6700 were hotter than E6400/6300 in general and it made sense in that the size of L2 contributes to operational temperatures. Then why are these 'native' 2MB/1MB L2 CPUs running so hot?

Edit: Wait, I missed Arkainium's posts. Maybe not all these E21x0s are crippled, after all! Incredible overclock there. Really, 3.60GHz for ~$85 can't be beat. That's surely faster than E6850 at its stock speed.

 

[PLUMMAN53]

Please post my results: E4300 with Thermaltake Typhoon cooler running at 3.3g on Gigabyte P35L with GSkill 6400. Vcore is 1.49 at 42C. FSB is 370 X 9.
Ram is stuck on 5 5 5 15 at 1110mhz at 2.1v.
750w Thermatake PS
85% Overclock so far.
CPU temp is around 42 C

 

[qliveur]

It just occurred to me to ask how many of you are using the BSEL mod.

 

[brencat]

BSEL mod is a workaround for P965 chipset mobos that don't have ability to change the strap from 800 to 1066 or 1333. Some boards by Abit (Quad GT) and DFI (965-S) had a strap selection feature making the mod unnecessary. On P35 boards like Abit IP35-E, the divider you pick will directly determine the strap -- either 1066 or 1333.

IMO, a 1333 mod is not advised with these chips in case CPU doesn't boot on stock volts at 333 FSB.

 

[SerpentRoyal]

Please Update (more details):

CPU: E4300 @ 3.47GHz
Multi x FSB: 9x386
Core voltage: 1.465V
MB: Abit IP35-E 12 BIOS
RAM speed/Memory divider: 483MHz (965 effective at 2.0V and 5-4-4-8-2T)/1:1.25. 1GB x 2 Kingston "N5" 1.8V DDR2 800 ValueRAM.
CPU cooler: Big Typhoon

 

[RocKKer]

Originally posted by: SerpentRoyal
Too much heat with the Freezer? 1.4V is a very mild oveclock with those C2D chips.

I was getting 66 - 67c (stressed) with the AC pro 7 and as5 at the higher clock and volt, I followed the directions on the AS to apply the thin strip right down the middle, did I apply right or there a better way?

 

[SerpentRoyal]

67C is fine under Orthos. No issue with the way you apply the AS5. Run at the higher voltage/core speed.

 

[RocKKer]

Originally posted by: brencat

Originally posted by: RocKKer
Please add my results:

E2180 Q712A452 @ 3.33GHz
9x370
1.4V
Abit IP35 v13 bios
AC Freezer 7 pro
Crucial Ballistix (2 x 1 GB) 1:1.20 DDR2 900 MHz 5-5-5-15 2.10v

Rockker, that's a nice result. If you don't mind, where and when did you purchase that chip?

Edit: NVM, I just realized all E2180 are MO stepping.

I agree with Serpent...with a better cooler you could probably try for higher OC if you're only at 1.4 vcore now.

I ordered the Abit and CPU and Crucial as an Mwave combo deal on Sep 21st.

Yes ur right I'm sure I could run higher with better cooling. However this is the wife's machine, she could care less how much I got it to OC! For her it just has to be runnin and I get nothing extra for a high FSB.

However this may turn out to be an OC test for a future rig for me.

 

[RocKKer]

Originally posted by: SerpentRoyal
67C is fine under Orthos. No issue with the way you apply the AS5. Run at the higher voltage/core speed.

SerpentRoyal - u r fast at replies. Ok on the higher temps...when is it dangerous would you say?

 

[SerpentRoyal]

I'd limit to 70C under Orthos Large mode. Wife's CPU will probably idle 90% of the time.

 

[A554SS1N]

Slighly OT: I'm throwing out ideas here, nothing is certain yet, but I'm possibly going to get a second PC, and for the processor, basically wanted something cheapish and be able to OC to 3.0Ghz for certain without stressing the motherboard - options include the E4x00 processors, or getting the E6550, but while the latter is quite a bit faster in some instances clock for clock, it's also pushing the board alot more IMO (FSB wise and possibly chispet heat wise) already, and I'd have to spend more to get a board that would have a high enough FSB limit to overclock it I assume - Ideally thinking of a half-decent board that's at the cheaper side of things. Of course if you guys still think maybe the E6550 is still worth going for, I'll probably go for that and OC it. Cheers.

 

[brencat]

Well...the Abit IP35-E that a bunch of us have here is pretty cheap and should allow you 425 - 450 FSB with minimal issue. The real question is are you comfortable shelling out almost $170 for the E6550 where you're limited to a 7x multiplier? For me, even at 450 x 7 (3.15ghz), that's not a good trade -- cost per clock. It's the same exact issue with the E6320 -- you have to depend totally on FSB clocking due to the 7x multiplier. And at that price point, you could buy an E6750 with the 8x multiplier for only $10 - $15 more and give yourself a shot at 3.6ghz running 450 x 8.

It really depends what you want it for. I recently came to the conclusion that I really want a Penryn quad, but also want something to play around with for the next few months. But if you only want 3.0 ghz, I'd say you're better off with the budget chips we're discussing in this thread.

 

[RocKKer]

Originally posted by: SerpentRoyal
I'd limit to 70C under Orthos Large mode. Wife's CPU will probably idle 90% of the time.

OK 70c sounds good.

I was thinking 60c was what I should shoot for. Before this I was used to OC older AMD's where 40c from the off chip thermistor (add about 20c to get on chip temp) was my target.

Runinng Orthos LM is probably the most that system will ever be stressed. I'll just leave it where it's at for a little cushion when the ambient gets higher, you know those summery days when she's stressing it with email!

SerpentRoyal I can't thank you enough for the help you have given me.

 

[lopri]

Like brencat says, I would definitely stay under 450FSB if to keep the rig for a long period of time. Minimal stress on memory as well. (I'd say 2.1V max w/o a fan and 2.2v w/ a fan) That's at least my personal opinion after dealing with quite a few intel boards and Micron sticks. The gains by stressing boards and memory are small but the chance of gradual degradation is disproportionately big.

 

[SerpentRoyal]

Originally posted by: brencat
Well...the Abit IP35-E that a bunch of us have here is pretty cheap and should allow you 425 - 450 FSB with minimal issue. The real question is are you comfortable shelling out almost $170 for the E6550 where you're limited to a 7x multiplier? For me, even at 450 x 7 (3.15ghz), that's not a good trade -- cost per clock. It's the same exact issue with the E6320 -- you have to depend totally on FSB clocking due to the 7x multiplier. And at that price point, you could buy an E6750 with the 8x multiplier for only $10 - $15 more and give yourself a shot at 3.6ghz running 450 x 8.

It really depends what you want it for. I recently came to the conclusion that I really want a Penryn quad, but also want something to play around with for the next few months. But if you only want 3.0 ghz, I'd say you're better off with the budget chips we're discussing in this thread.

And you'll need very good overclocking RAMs to do 460MHz (3.2GHz core speed). E2160/E2180/E4300/E4400/E4500 can easily hit 3.2GHz with DDR2 667 RAMs ($15/GB HP). Core speed is KING.

 

[SerpentRoyal]

Originally posted by: RocKKer

Originally posted by: SerpentRoyal
I'd limit to 70C under Orthos Large mode. Wife's CPU will probably idle 90% of the time.

OK 70c sounds good.

I was thinking 60c was what I should shoot for. Before this I was used to OC older AMD's where 40c from the off chip thermistor (add about 20c to get on chip temp) was my target.

Runinng Orthos LM is probably the most that system will ever be stressed. I'll just leave it where it's at for a little cushion when the ambient gets higher, you know those summery days when she's stressing it with email!

SerpentRoyal I can't thank you enough for the help you have given me.

No prob. Note that the older AMD chips report AVERAGE CPU die temperature. CoreTemp reports CORE CPU temperature, which is 15C hotter than the average CPU temperature.

 

[Avalon]

Originally posted by: Arkainium

Originally posted by: Avalon
Did you have to modify your TR Ultra Extreme to get it to seat properly on the bare core, or did it just work? I've got an E2140 and a TR Ultra-120, and am seeing some high temps...don't really care to lap, but if I can just hack off the IHS, I wouldn't mind giving that a shot if I didn't have to do any further modifications.

The problem is the little swinging latch on the motherboard that locks the cpu to the socket sits too high to get good contact between the core and the heat sink. So I took that latch off but I still wasn't making good contact because the screws that come with the Ultra-120 Extreme don't go down all the way. So I used some washer-like things to make it tighter.

At first it wouldn't post, so I thought I messed up. But the problem is without the latch the cpu might not be secure so you need to be careful when you mount the heatsink without it. The latch comes off really easy, so it's just a matter of getting the heat sink on tight enough to hold the chip in place.

Eesh, sounds like a bit much for me. Kudos on getting that to work, though. I think I'll just settle in on my current OC, since it's at low voltage.

 

[Uhtrinity]

Initial results:

E2160 1.8 Ghz @ 3.4 Ghz (9 X 377), 1.4volts
2 GB of Wintec AMPX DDR 800 @ 754, haven't changed memory timing from defaults yet.
Gigabyte GA-P35-DS3L
Coolermaster Aquagate water cooling
38 - 40c under full load running Prime 95 on each core, 30ish idle.

Anyone have any ideas what bus speed this board is capable of? Running the FSB / northbridge at stock voltage atm.

Essentially the same system as in my sig with the new MB, CPU, and memory. Upgrade cost - $250

Edit:

One Core failed P95 over night, bumped the core voltage upto 1.425v, retesting

 

[Avalon]

Originally posted by: Uhtrinity
Initial results:

E2160 1.8 Ghz @ 3.4 Ghz (9 X 377), 1.4volts
2 GB of Wintec AMPX DDR 800 @ 754, haven't changed memory timing from defaults yet.
Gigabyte GA-P35-DS3L
Coolermaster Aquagate water cooling
38 - 40c under full load running Prime 95 on each core, 30ish idle.

Anyone have any ideas what bus speed this board is capable of? Running the FSB / northbridge at stock voltage atm.

Essentially the same system as in my sig with the new MB, CPU, and memory. Upgrade cost - $250

Edit:

One Core failed P95 over night, bumped the core voltage upto 1.425v, retesting

Ok, now I definitely know something isn't right with my temps, since you have the same motherboard I do. Here are my overclock recordings I made at each step I achieved.

E2140 SLA3J Q706A148
1600mhz @ 200x8 1.325v - (33/32C)/(46/46C)/76F (Stock Intel Cooler)
2130mhz @ 266x8 1.325v - (27/24C)/(43/41C)/76F (TR Ultra-120)
2400mhz @ 300x8 1.325v - (29/27C)/(44/43C)/76F (TR Ultra-120)
2664mhz @ 333x8 1.325v - (29/27C)/(46/46C)/76F (TR Ultra-120)
2800mhz @ 350x8 1.325v - (29/27C)/(48/48C)/76F (TR Ultra-120)
2900mhz @ 363x8 1.35v - (29/27C)/(51/51C)/76F (TR Ultra-120)

First temperature bracket is the idle of core 0 and core 1, respectively, using Speedfan.

The second temperature bracket is the Orthos load of core 0 and core 1, respectively, again using Speedfan.

The loose temperature reading at the end is just my ambient temperature inside the house.

As you can see, despite me not having a 1600mhz temperature bench of my TR Ultra-120, I still gained relatively little in temperature reduction when fully loaded.

At my final overclock, I'm running 11C higher than you fully loaded using less voltage and running 500mhz less. If I add 15C to approximate Core Temp, I'm also running waaaay hotter than AT did on their max overclock load of their Core 2 Extreme.

I'm pretty positive I installed the heatsink correctly...I've certainly installed a fair share of 3rd party heatsinks, and the installation of the Ultra-120 is a no brainer...you screw it in. Maybe my IHS is bad?