Please visit donation page to help the project cover running costs for this month

Sponsored by:

Hosted at and sponsored by Rackspace.

Join PrimeGrid

Returning Participants


Leader Boards



1) Message boards : Number crunching : PrimeGrid is looking for a new home (Message 109875)
Posted 199 days ago by VortacProject donor
PrimeGrid is accepting both BTC and GRC donations (and PayPal, of course):
2) Message boards : Generalized Fermat Prime Search : Invalid GFN WUs when validating ATI against OCLcuda (Message 94370)
Posted 699 days ago by VortacProject donor
Very rarely I get an invalid Genefer result, but when I do, it's always against two OCLcuda wingmen and it's always GFN17MEGA. Could this be a validation issue?

I used to run much more demanding GFN-WR workunits before (on the same machine) and I never got an invalid result. It's seems strange to me that GFN17MEGA is producing errors, if GFN-WR hasn't.
3) Message boards : Number crunching : Oops! Lesson learned about UPS and active PFC power supplies (Message 74053)
Posted 1478 days ago by VortacProject donor
Regarding wattage readouts, what UPS model is that from?

PowerChute screenshot is for Back UPS Pro 1500 (BR1500GI).
Apcupsd screenshot is for Back UPS RS 800 (BR800I).

Eight years ago, I also used APC Back UPS CS 500 (CS series - leaning towards the low-end) and it also displayed actual power consumption through PowerChute. So it came as a surprise to me that some APC models don't support that (basic) feature.

Anyway an interesing topic, since more and more people are equipping their BOINC machines with UPSs, especially if the machine is unattended. It's more complex than it looks, choosing an UPS for such a power-hungry machine.
4) Message boards : Number crunching : Oops! Lesson learned about UPS and active PFC power supplies (Message 74050)
Posted 1478 days ago by VortacProject donor
Unless it's changed, their consumer UPSs don't show the wattage actually being used. Only their higher end UPSs do. Certainly, none of the APC UPSs I've ever owned at home do. The ones at work (usually Smart-UPS) do. I'm not sure about Back-UPS Pro.

Well, perhaps we have a misunderstanding regarding the "wattage actually being used" so let's have screenshot of the APC PowerChute software:
"Your battery backup is currently providing 553 watts of power"

Even open-source software like apcupsd can provide the wattage information on APC UPSs. My other UPS is monitored via apcusd, since you cannot monitor more than one UPS via PowerChute Personal Edition. Here is the screenshot of apcupsd software:

Load percentage (LOADPCT) is 67%. Since that's a 540W unit, actual load is 362W. Total load of my system is then 553+362=915W, a lot of power - provided by two Seasonic X-Series ActivePFC PSUs running on two stepped sine-wave output, consumer-grade APC UPSs.

APC has addressed the issue of running Active PFC PSUs on their line of consumer, stepped sine-wave output UPSs. Basically, their advice in such cases is to size the UPS according to the maximum rated output of the Active PFC power supply (due to potential power rushes), not according to the peak power consumption of the PC. I haven't experienced such power rushes with Seasonic PSUs, but I did in the past, with a cheaper Chieftech unit (GPS-500AB-A, also with Active PFC). That power-rush phenomenon with Chieftec PSUs is also described in this Xbit labs article:

Regarding the APC advice, you can find it in their FAQ knowledge base. Since a direct link can't be provided, I will copy-paste the most important part:

A computer’s power supply may also be subjected to a period of inrush, while the UPS is changing state (switching from utility power to battery power and back). Back-UPS and Smart-UPS SCs may experience up to an 8ms transfer time during this period. This is just long enough to remove power from the PFC power supply, resulting in a momentary inrush of the PFC. Once the UPS changes states from "Online" (passing utility power) to "Onbattery" (passing power from the UPS's internal battery), the momentary inrush from the attached equipment subjects the UPS to the On battery power supply’s maximum power draw, resulting in a potential Overload condition or dropped load.
An Energy Star 4.0 compliant power supply has to be more than 80% efficient. For example, if a attached power supply is delivering 600W output power, its ‘input’ power can be as high as 750W. .
This ‘input’ power should be the basis for sizing the UPS, so as not to Overload the UPS. This can be calculated by taking the PFC power supply’s rated output power and multiplying it by 1.25
5) Message boards : Number crunching : Oops! Lesson learned about UPS and active PFC power supplies (Message 74024)
Posted 1478 days ago by VortacProject donor
Cyberpower also has a Windows 7 desktop gadget that shows various things, including watts. As far as I know, the consumer level APC UPSs do not provide that information. They can tell you the battery state, but not the power being used.

That's not true, APC PowerChute software shows the power being used and actual input voltage, even for consumer level UPSs - personally tested so far on APC Back UPS Pro and Back UPS RS & CS. I got my first APC UPS more than ten years ago and it had those features even then.

Also, I currently have an APC UPS with step-approximated sine-wave output (Back-UPS Pro 1500, BR1500GI) and it works fine with my Active PFC PSU (Seasonic X-650,SS-650KM) - the battery kicks in normally every time the power goes out or if there is electrical noise.
6) Message boards : Number crunching : Year of the Horse - Stallion Edition (Message 72775)
Posted 1514 days ago by VortacProject donor
It's Nvidia GPUs that are particularly sensitive to overclocking when running Genefer. CPUs (and possibly ATI GPUs) are, as far as I know, just "normally" sensitive to overclocking.

I can confirm this. 7970s overclock reasonably well with GNFWR. However, it's still more challenging than other BOINC projects. For example, GNFWR needs 1.250V for total stability (three watercooled 7970s, running on 1200 MHz). DistrRTgen needed 1.225V.
7) Message boards : Number crunching : Year of the Horse - Stallion Edition (Message 72684)
Posted 1516 days ago by VortacProject donor
Thank you both for the answers. Yes, glancing at top hosts now, Titan (OpenCL) manages one GNFWR workunit in about 205 000 seconds and 780 Ti in ~262 000 seconds. That makes 780 Ti 28% percent slower which certainly looks more realistic, however still very impressive performance from 780 Ti, considering it's primarily a gaming card (with seriously handicapped FP64 performance).
8) Message boards : Number crunching : Year of the Horse - Stallion Edition (Message 72654)
Posted 1516 days ago by VortacProject donor
I have a question regarding Nvidia GPUs performance in GNFWR. No.1 on "Top GPU models" is of course TITAN (1.000) but very close second is 780 Ti (0.917). Now that's puzzling to me, because 780 Ti double precision performance is capped at 1/24 FP32, so TITAN should be (in theory) approximately 8x faster, since its powerful FP64 is 1/3 FP32. But according to Primegrid, the difference is almost negligible! Does GNFWR also depend on FP32 performance? Or is there another explanation?
9) Message boards : News : Fix for screen lag when running GPU (Message 72176)
Posted 1530 days ago by VortacProject donor
In IE, it's Tools -> Internet Options -> Advanced -> "Use software rendering instead of GPU rendering" (on the very top of all advanced options)
10) Message boards : Number crunching : Year of the Horse - Stallion Edition (Message 72135)
Posted 1531 days ago by VortacProject donor
Thanks for quick explanation.
Reducing clocks by 25 MHz :)

Next 10 posts
[Return to PrimeGrid main page]
Copyright © 2005 - 2018 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 3.31, 3.35, 3.05
Generated 22 Mar 2018 | 2:57:03 UTC