| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 123
|
|
| Author |
|
|
Hype
Cruncher Germany Joined: Nov 18, 2011 Post Count: 43 Status: Offline Project Badges:
|
There is no performance loss through undervolting and it can save around 5-20% of power depending on the CPU.
----------------------------------------I am still testing with my Ryzen 1700 and so far I'm not that impressed, I thought it would use less power. After powering it on for the first time without changing anything in the BIOS I had 140 watts under full load with BOINC. After some undervolting and changing other settings I am now at ~ 115 watts. This is double the usage of my E3-1245 v2, and a bit more than double PPD. I'll now set up Linux Mint on both and let them run for a few days. This is not 100% accurate as the Ryzen system has got a GTX 780 in it, the E3 didn't have a GPU, as it has got an iGPU. Tom's hardware although says that the GTX780 only has 8 watts idle use, so it wouldn't be much less without it. My i5 4670k system (with the same GTX780) had 70 watts under full load. So I need to do a bit more testing. I don't think the mainboard takes 50 watts more than the old one, so it can be only the CPU. This proves that TDP has got absolutely nothing to do with actual power consumption, it just tells how much heat the CPU generates. The Ryzen 1700 has 65W TDP, the E3 77W. The E3 sits at around 60°C, the Ryzen a bit lower at 50-55°C, so that makes sense. So far it looks like the Ryzen 1700 takes around 40-50 watts more than my E3. It is more efficient, just not as much more as I thought it would be. I am impressed by the stock AMD cooler btw, it's very quiet and the temps are very low. Looks like it will be the first stock cooler I won't replace :-) ![]() [Edit 4 times, last edit by RoundFour at Dec 6, 2017 12:43:23 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I never even used my stock cooler. It had good reviews, but i had already picked up an aftermarket and stuck with that one. Shame that the 1700 doesnt seem to undervolt very well. I guess the guys on overclock.net were correct.
|
||
|
|
wolfman1360
Senior Cruncher Canada Joined: Jan 17, 2016 Post Count: 176 Status: Offline Project Badges:
|
What are you finding average PPD on your ryzen with windows? Or do you run linux exclusively?
----------------------------------------Thanks!
Crunching for the betterment of human kind and the canines who will always be our best friends.
AWOU! |
||
|
|
Hype
Cruncher Germany Joined: Nov 18, 2011 Post Count: 43 Status: Offline Project Badges:
|
At the moment I'm running Linux, I had Windows running for only 5 days.
----------------------------------------In that 5 days Windows had around 76k PPD, but 5 days is not enough for a proper average number. Linux so far has 84k PPD, so around 10% more. ![]() |
||
|
|
Hype
Cruncher Germany Joined: Nov 18, 2011 Post Count: 43 Status: Offline Project Badges:
|
Alright, so I compared both my systems over 12 days and this is the result:
----------------------------------------![]() Both run Linux Mint 18.3. The E3-1245 v2 does 41k PPD, the Ryzen 1700 88k. The Ryzen 1700 requires 73% more power, but does 113% more points, so it's more efficient :-) I'll try to get the power usage lower over christmas. Today I'll switch the Ryzen over to Windows to see how big the difference in PPD is. ![]() [Edit 1 times, last edit by RoundFour at Dec 19, 2017 4:32:03 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Interesting comparison RoundFour!
Could you share more details about your systems? - what frequencies do the CPUs run at? 3,2 GHz for the Ryzen and 3,6 GHz for the Xeon (=all core turbo boost)? - which power supply is used? (wattage and 80+ bronze/silver/gold whatever specification) - does Linux run from HDD, SSD or USB stick? Any other HDDs, CD/DVD drives or whatever installed? For the comparison between Linux/Windows: The result will be very dependent on projects crunched. I would assume that the difference for VINA projects (SCC, Zika, OET, FAH1) will be largest and smaller for the other projects. Best strategy might be to compare Linux/Windows for every project separately. Listing all valid workunits in "Result Status" on the WCG site for the PC/OS to be examined and calculating granted credit / cpu time for a sufficiently large number of work units and looking at the average should work well. |
||
|
|
hchc
Veteran Cruncher USA Joined: Aug 15, 2006 Post Count: 865 Status: Offline Project Badges:
|
A standardized unit of comparison such as PPD/Watt is sometimes easier:
----------------------------------------E3-1245v2 Efficiency = 41,561/71 = 585 PPD/Watt Ryzen 1700 Efficiency = 88,441/123 = 719 PPD/Watt
[Edit 1 times, last edit by hchc at Dec 20, 2017 2:57:59 AM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Alright, so I compared both my systems over 12 days and this is the result: ![]() Both run Linux Mint 18.3. The E3-1245 v2 does 41k PPD, the Ryzen 1700 88k. The Ryzen 1700 requires 73% more power, but does 113% more points, so it's more efficient :-) I'll try to get the power usage lower over christmas. Today I'll switch the Ryzen over to Windows to see how big the difference in PPD is. Interested to see how much lower you can get the wattage on the 1700. I'll keep an eye out. I know you initially said it didn't drop much during initial testing, but hopeful to squeeze a little more efficiency out of mine as well. |
||
|
|
Hype
Cruncher Germany Joined: Nov 18, 2011 Post Count: 43 Status: Offline Project Badges:
|
@hchc: Thanks, that makes sense.
----------------------------------------@Sheridon: Sure, here are more details: E3-1245 v2 @ 3,4 Ghz on Mini-ITX board 350W 80+ Bronze 84% 2.5" Toshiba HDD, no other drives, disabled everything in BIOS which is not needed (audio etc.), cpu and ram undervolted to hell Ryzen 1700 @ 3,2 Ghz on ATX board 480W 80+ Gold, 90% SSD, no other drives (additional HDD requires only ~ 2 watts in idle) This system has a GTX 780 in it, not sure about idle usage, Google says around 8-10 watts. ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Tried to get the energy usage down on my Dell Optiplex with i7-3770 today.
- runs Linux from USB stick and no other drives, so no further improvements here - removed RAM, has only 4 GB left now, which is enough for WCG - left case open and switched system fan off - deactivated SATA, all USB ports not needed and PCI in Bios - system has a 250W gold rated PSU, so this should be reasonably efficient - undervolting is unfortunately not possible with the DELL motherboard Power when running WCG on all threads: 83 W with Turbo Boost activated (3,7 GHz) 69 W with Turbo deactivated (3,4 GHz) Will leave it at 3,4 GHZ, as 17% less power but only 9% less clock rate seems to be worth it. Seems pretty efficient for such an old system, however not as efficient as a Ryzen. Looking forward to build one in January! |
||
|
|
|