Core 2 Duo, much lower power consumption with older chipsets
Posted: Wed Aug 16, 2006 9:08 am
I was disappointed that the over-clocking test that I performed with C2D using an Asus 975X based motherboard only allowed an idle power consumption of 88W at minimum. This was with Speedstep enabled and using CrsystalCPUID didn’t allow under-volting and neither did the BIOS.
I received an ASRock ConRoe945G-DVI today and preliminary tests show it idling at 58W when using the onboard graphics in combination with the same hardware as the previous test. The one other difference between the two tests is that the initial one used 3 Nexus 120mm fans at 600 rpm, whereas the current test is using 1 Nexus 120mm at 1100 rpm. That’s not a significant difference and I might try and quantify it later.
Comparing the two tests gives this data;
Chipset, VGA, GHz, VCore (at load), Watts (Idle, CPU Burn-In, Prime95),
975X / 6200TC / 2.13 GHz / 1.150V / 088 102 112 Watts
945G / GMA950/ 2.13 GHz / 1.196V / 058 074 085 Watts
945G / 6200TC / 2.13 GHz / 1.196V / 066 090 101 Watts
A 30W difference at idle when using the onboard graphics is excellent. But I’m really surprised by the 22W difference when using the same VGA card in both boards.
I’m not sure why the 6200TC consumes so much power in the ASRock when the system is under load but with no GPU load! That seems a strange anomaly!!! Anyone?
Both the idle wattages are at 1.15V as Speedstep was enabled; the Vcore shown is for the two tests at load. The Asus 975X test failed Prime95 so would have needed to have run at a slightly higher VCore, which means the wattage under load would have also been slightly higher.
With the ASRock board the Vcore was set to 1.196V using CrystalCPUID although Speedfan reported voltage as being 1.18V. I haven’t tried dropping the voltage lower than this yet at 2.13 GHz to see if it’s still stable, although I will later in the week.
The ASRock was dual Prime95 stable for 1 hour and the VCore looked very stable, more so than the Asus board which cost three times the price.
The other good thing about these inexpensive C2D boards is that when Intel releases the E4300 towards the end of 06 beginning of 07, because it will have a FSB of 200 compared to 266 for current C2Ds, they will over-clock really well even on older chipset boards which are typically only reaching ~300 FSB.
The E4300 has a multiplier of 9 for 9 * 200 = 1.8 GHz. So running it at 9 * 300 = 2.7 GHz. Not bad for a $130 chip on a $50 motherboard.
It’ll be interesting to see if the lower Speedstep multiplier is also set for 6 on the E4300 chips. That would give it a lower Speedstep idle speed of 6 * 200 = 1.2 GHz, quite a bit less than 1.6 GHz of current C2Ds. If the voltage is still restricted to a minimum of 1.15V, then it shouldn’t make any significant difference to the idle power consumption.
It’s preferable for the E4300 to use a minimum multiplier of 6 as when over-clocked to FSB = 300, it would have a Speedstep range of 1.8 to 2.7 GHz, which is easily achievable using the seemingly fixed C2D Speedstep voltage range of 1.15 to 1.325V.
In fact the ASRock wouldn’t go any higher than 1.29V according to Speedstep, as opposed to 1.325V with the Asus board.
According to Intel’s spec sheet for the E6400 that I was using, it has a voltage range of 0.85 – 1.3525V, as do all the E6x00 chips. This seems unusually low unless the Spec is for all C2D chips including the mobile ULV parts. It’s not really feasible for a C2D to idle at 1.6 GHz @ 0.85V, let’s be honest.
The ASRock board retails for under £50 in the UK and supports DVI and dual monitor configurations out of the box. It also sports 4x SATAII ports, 1x PATA, 4x DDR2-533/667 slots, 1x PCIe x16, 2x PCI, Gigabit NIC..
I received an ASRock ConRoe945G-DVI today and preliminary tests show it idling at 58W when using the onboard graphics in combination with the same hardware as the previous test. The one other difference between the two tests is that the initial one used 3 Nexus 120mm fans at 600 rpm, whereas the current test is using 1 Nexus 120mm at 1100 rpm. That’s not a significant difference and I might try and quantify it later.
Comparing the two tests gives this data;
Chipset, VGA, GHz, VCore (at load), Watts (Idle, CPU Burn-In, Prime95),
975X / 6200TC / 2.13 GHz / 1.150V / 088 102 112 Watts
945G / GMA950/ 2.13 GHz / 1.196V / 058 074 085 Watts
945G / 6200TC / 2.13 GHz / 1.196V / 066 090 101 Watts
A 30W difference at idle when using the onboard graphics is excellent. But I’m really surprised by the 22W difference when using the same VGA card in both boards.
I’m not sure why the 6200TC consumes so much power in the ASRock when the system is under load but with no GPU load! That seems a strange anomaly!!! Anyone?
Both the idle wattages are at 1.15V as Speedstep was enabled; the Vcore shown is for the two tests at load. The Asus 975X test failed Prime95 so would have needed to have run at a slightly higher VCore, which means the wattage under load would have also been slightly higher.
With the ASRock board the Vcore was set to 1.196V using CrystalCPUID although Speedfan reported voltage as being 1.18V. I haven’t tried dropping the voltage lower than this yet at 2.13 GHz to see if it’s still stable, although I will later in the week.
The ASRock was dual Prime95 stable for 1 hour and the VCore looked very stable, more so than the Asus board which cost three times the price.
The other good thing about these inexpensive C2D boards is that when Intel releases the E4300 towards the end of 06 beginning of 07, because it will have a FSB of 200 compared to 266 for current C2Ds, they will over-clock really well even on older chipset boards which are typically only reaching ~300 FSB.
The E4300 has a multiplier of 9 for 9 * 200 = 1.8 GHz. So running it at 9 * 300 = 2.7 GHz. Not bad for a $130 chip on a $50 motherboard.
It’ll be interesting to see if the lower Speedstep multiplier is also set for 6 on the E4300 chips. That would give it a lower Speedstep idle speed of 6 * 200 = 1.2 GHz, quite a bit less than 1.6 GHz of current C2Ds. If the voltage is still restricted to a minimum of 1.15V, then it shouldn’t make any significant difference to the idle power consumption.
It’s preferable for the E4300 to use a minimum multiplier of 6 as when over-clocked to FSB = 300, it would have a Speedstep range of 1.8 to 2.7 GHz, which is easily achievable using the seemingly fixed C2D Speedstep voltage range of 1.15 to 1.325V.
In fact the ASRock wouldn’t go any higher than 1.29V according to Speedstep, as opposed to 1.325V with the Asus board.
According to Intel’s spec sheet for the E6400 that I was using, it has a voltage range of 0.85 – 1.3525V, as do all the E6x00 chips. This seems unusually low unless the Spec is for all C2D chips including the mobile ULV parts. It’s not really feasible for a C2D to idle at 1.6 GHz @ 0.85V, let’s be honest.
The ASRock board retails for under £50 in the UK and supports DVI and dual monitor configurations out of the box. It also sports 4x SATAII ports, 1x PATA, 4x DDR2-533/667 slots, 1x PCIe x16, 2x PCI, Gigabit NIC..