Kill-a-watt

KayBee

Active member
Feb 24, 2020
135
102
43
Hi All, this is not new information probably. I recently measured the wattage my 300MHz R5k and 400MHz R12k Octane use when running typical uses. The Kill-a-watt device is accurate enough for me, I wanted to see what my SGIs use vs. my Cylinder Mac Pro, and my sons Gaming rig with big RTX card.
O2 (SCSI2SD, 128MB RAM): ~64 Watts when compiling
Octane (SCSI2SD, SI graphics, 2GB RAM: ~210 Watts when compiling
2013 12-core D700 Mac Pro, 48GB RAM: ~140 Watts (up to 230 when Minecraft with settings cranked)
Hexacore 3.5GHz i7, 32GB RAM, RTX 3060ti: 120 Watts (up to 280 when Minecraft...)

The difference in SGIs is not surprising I suppose, the Cherokee supply weighs as much as the O2.

Well this is about as far from scientific as can be, but I live in an old house with old power, and the wrong combo of machines on at the same time causes the breaker to trip. I have all machines in UPS, but its a pain to go outside and reset. So I have the machines on different circuits now and all is good.

I'd be curious what other machines use, if you feel like posting.

Cheers All.

KB

-FYI Edited to remove unfinished sentence about the Mac-
 
Last edited:
  • Like
Reactions: rbz

jupo

New member
Jul 9, 2019
13
4
3
44
Tucson, AZ
My Octane with the Lucent PSU, SSE and SI video, and two SCSI drives pulls about 330W. There's virtually no difference between idle and very heavy load.

For comparison, my iMac G4 at idle with the screen off registers about 20W. :)
 

Elf

Storybook / Retired, ex-staff
Feb 4, 2019
792
252
63
Mountain West (US)
Looks about right! I should also note that you may want to measure Volt-Amps (VA: that is, apparent power) or even just straight Amps (A), as well as Watts (W: true power). I think the Kill-a-Watt has a setting for this?

While a wattage measurement indicates power actually consumed and turned into work, there is a reactive power component that does not perform any work but still draws current through the breaker, which is not represented in that measurement but will nonetheless contribute to circuit loading and eventual breaker tripping.

{True Power (W)} = {Power Factor} * {Apparent Power (VA)}, so supplies with a low power factor (and the power factor of some older SGIs is abysmally low) can draw significantly more current than their wattage might otherwise indicate. Newer ones that are power factor corrected (PFC) like the Octane supply aren't too bad, so VA will be closer to W. However for older non-PFC supplies this is not nearly the case. This will be of particular concern for people running old, heavy draw items such as desksides.

Most breakers have an 80% trip rating, so a 15A breaker will eventually trip if continually loaded over 12A, though this might take hours. Just remember: breakers trip on current, which is only partially related to wattage. So if your primary concern is circuit loading and breaker tripping, measure current and not just wattage.
 
Last edited:
  • Like
Reactions: jenna64bit and rbz

KayBee

Active member
Feb 24, 2020
135
102
43
My Octane with the Lucent PSU, SSE and SI video, and two SCSI drives pulls about 330W. There's virtually no difference between idle and very heavy load.

For comparison, my iMac G4 at idle with the screen off registers about 20W. :)
Thanks Jupo, that gives me a ballpark for a more beefed up Octane.
Cheers,
KB
 

KayBee

Active member
Feb 24, 2020
135
102
43
Looks about right! I should also note that you may want to measure Volt-Amps (VA: that is, apparent power) or even just straight Amps (A), as well as Watts (W: true power). I think the Kill-a-Watt has a setting for this?

While a wattage measurement indicates power actually consumed and turned into work, there is a reactive power component that does not perform any work but still draws current through the breaker, which is not represented in that measurement but will nonetheless contribute to circuit loading and eventual breaker tripping.

{True Power (W)} = {Power Factor} * {Apparent Power (VA)}, so supplies with a low power factor (and the power factor of some older SGIs is abysmally low) can draw significantly more current than their wattage might otherwise indicate. Newer ones that are power factor corrected (PFC) like the Octane supply aren't too bad, so VA will be closer to W. However for older non-PFC supplies this is not nearly the case. This will be of particular concern for people running old, heavy draw items such as desksides.

Most breakers have an 80% trip rating, so a 15A breaker will eventually trip if continually loaded over 12A, though this might take hours. Just remember: breakers trip on current, which is only partially related to wattage. So if your primary concern is circuit loading and breaker tripping, measure current and not just wattage.
Interesting! Thank you for the explanation. A nice science class for me. I’ll try VoltAmps and Amps as well, since the “Kill-A-Watt device can read them.
OT I suppose but I wish brand new O2 PSUs were available. I would say the same for Octane PSUs but I can imagine what they would cost.
 
  • Like
Reactions: Elf

drmadison

Member
Jun 30, 2020
33
20
8
I had done some similar measurements.
My Octane2 in its older spec (single 550mhz R14k, V12+DCD, SSD via ACARD adapter) sits in the 245-260w range. That's a bit higher since bumping up to the dual 600mhz cpus as would be expected though I can't find anywhere I've written down measurements. Also, it measured about 4w when powered down.

Some other measurements I did that WERE documented:

SGI O2 180mhz R5000 old 8gig scsi HDD, AV card, no CDrom, spiked to 95W on boot, 76W idle, 83W under load (heavy disc access included), power factor of 0.98
SGI O2 300mhz RM5200 old 8gig scsi HDD, AV card, no CDrom, spiked to 95W on boot, 77W idle, 83W under cpu load (heavy disc access included), 88W under heavy cpu+graphics load, power factor of 0.98

Swapping out some of the ram, changing the power supply fan, and ditching that old 8-gig SCSI drive for a SCSI2SD (externally powered, not included in this measurement) dropped the 300mhz RM5200 to sit around 66W in use.
 

KayBee

Active member
Feb 24, 2020
135
102
43
Thanks for adding that drmadison, validates my O2 measurement. Using the Octane is a snappier experience for my uses, but I can't help but keep firing up the O2 and using it for developement.
 

jenna64bit

Administrator
Apr 18, 2020
86
23
8
Nice! I am not at home so can't give you exact measures but playing around with mine a while back I found measurable wattage changes with MXE vs SE. Further, I found the later faster CPUs (which are die shrunk) are lower power consumption. Haven't tested with my V12 yet though. Around 320 sounds around right. My O2 is an R12K and still I think was around 96? That's all completely off the top of my head.
 
  • Like
Reactions: Elf

KayBee

Active member
Feb 24, 2020
135
102
43
Forehead-slap! Good Point Jenna, I didn't think of that. I have an R14k 600 Octane cpu but I have not been using it. Now I need to try it.
 

khral

New member
Aug 5, 2021
3
6
3
Thailand
Add more data from my SGIs

Indigo2 (R4400 250Mhz, 384MB RAM, GU1-Extreme, 72GB 10K HDD and ZYTEC 385W (P/N 060-0021-001) PSU)
- Standby, 11.4W
- Idle, 156W
- Running PowerFlip + SolidView + Ideas demos (in large window), 164W Power factor of 0.56

Tezro (Quad R16000 800 MHz (4MB L2), 3GB RAM, V12 with DCD-2, two 72GB 15K HDD and Cougar GX 1050W PSU)
My Tezro is using modern ATX PSU, so the power usage may not represent a Tezro with original PSU. Also, my Tezro have usual Discreet Flame cards set plus a PCI-X SAS/SATA Card and will use more power than a bare bone Tezro.
- Standby, 4.5W
- Idle, about 308W
- Playing Quake 3, 320-324W
- Maya 6.5 mentalray rendering (seems not use all 4 CPUs according to gr_osview), 306W
- Compiling Blender 2.49b with sgug-rse (using 4 threads), 305-308W Power factor of 0.96

Seeing O2 power usage make me want one, just to reduce my electric bill a bit.
 
  • Like
Reactions: Elf and flexion

WDB

New member
Aug 12, 2021
5
1
3
Southern California
My Octane with the Lucent PSU, SSE and SI video, and two SCSI drives pulls about 330W. There's virtually no difference between idle and very heavy load.

For comparison, my iMac G4 at idle with the screen off registers about 20W. :)
I just got an O2, my first SGI computer, seeing the power draw of your Octane makes me question if I want an Octane now. I have a high power HP Z640 that I use for graphics stuff, I limit its use due to how much power it uses, in Blender rendering with all 20 CPU cores it pulls 375 watt on my Kill-A-Watt. Electricity is expensive here and I can play around with an O2 for a lot longer without driving my Electric bill up too much.
 

About us

  • Silicon Graphics User Group (SGUG) is a community for users, developers, and admirers of Silicon Graphics (SGI) products. We aim to be a friendly hobbyist community for discussing all aspects of SGIs, including use, software development, the IRIX Operating System, and troubleshooting, as well as facilitating hardware exchange.

User Menu