386SX wrote:

I usually buy too the cheap version of a good quality psu. What do you think about FSP psu?

FSP Group is the OEM supplier that HP used for all of their rack-mount server machines at least back from 2002 - 2008. Source: I worked in IT at the time and often had the covers off to replace cpu/ram and always saw FSP Group on the power units. Sometimes hot-swapped the power units in some chassis' and had spares on a pallet.

So if they're good enough for HP mission-critical servers, they should be good enough for your home computers. If you search on google for PDF documents there's lists of FSP group units (FOR OEM) that are actually 80+ multi-tier of different color segments too. But FSP doesn't advertise this on the unit it's self. Unlike say EVGA or others "For home consumers" that flash it all over their units and boxes and stuff. With FSP you just have to look up the model # and know what you're buying. Which if you do some research can be a big benefit because without the labeling, sometimes people on ebay list gold and platinum tier FSP power supplies for cheaper than most counter-parts from the likes of Corsair, EVGA, ETC, just because they don't realize what they have.

There are some FSP units that flash the 80+ or 85+ on the side.. but there are also some nice 80+ oem ones that were inside like dell or hp desktops that are still 80+ but with no logo sometimes.

Tetrium wrote:

I don't think it matters much if your rig is under 200W or so when measuring from the wall. Even my Phenom II never seemed to consume that much, but that's also because I stick to mid-range graphics (and I'm not using Seasonic or another absolutely top-brand PSU, I've always gone for the next best thing like FSP) and I'm more of a "playable framerates" guy and don't even have any HD displays here. I actually tend to disable all the extra stuff like smoke because to me that's just an extra annoyance I don't want to have to deal with 😜

In my case, I don't use/need my machines for (at least 3D) gaming or heavy programs, so it becomes easier to get low power figures from them. 😜

I would just pull the GeForce from the P4 above and use the IGP if, 1) doing this netted more than just ~3-4w of savings at idle/load; 2) the motherboard has a DVI port, which it doesn't.

386SX wrote:

By the way... Sempron dual 1,45Ghz, Radeon R5 with on demand power load: 22W in idle oh yeah. 😁

Very nice 😁

kithylin wrote:

FSP Group is the OEM supplier that HP used for all of their rack-mount server machines at least back from 2002 - 2008. Source: I […]
Show full quote

386SX wrote:

I usually buy too the cheap version of a good quality psu. What do you think about FSP psu?

FSP Group is the OEM supplier that HP used for all of their rack-mount server machines at least back from 2002 - 2008. Source: I worked in IT at the time and often had the covers off to replace cpu/ram and always saw FSP Group on the power units. Sometimes hot-swapped the power units in some chassis' and had spares on a pallet.

So if they're good enough for HP mission-critical servers, they should be good enough for your home computers. If you search on google for PDF documents there's lists of FSP group units (FOR OEM) that are actually 80+ multi-tier of different color segments too. But FSP doesn't advertise this on the unit it's self. Unlike say EVGA or others "For home consumers" that flash it all over their units and boxes and stuff. With FSP you just have to look up the model # and know what you're buying. Which if you do some research can be a big benefit because without the labeling, sometimes people on ebay list gold and platinum tier FSP power supplies for cheaper than most counter-parts from the likes of Corsair, EVGA, ETC, just because they don't realize what they have.

There are some FSP units that flash the 80+ or 85+ on the side.. but there are also some nice 80+ oem ones that were inside like dell or hp desktops that are still 80+ but with no logo sometimes.

Thanks, next time I'll surely check the efficiency and gold,bronze etc.. standard. 😀 FSP is easier to find in shops, for other brands need to order it.

386SX wrote:

There are some FSP units that flash the 80+ or 85+ on the side.. but there are also some nice 80+ oem ones that were inside like dell or hp desktops that are still 80+ but with no logo sometimes.

Thanks, next time I'll surely check the efficiency and gold,bronze etc.. standard. 😀 FSP is easier to find in shops, for other brands need to order it.[/quote]

Just.. if you find a FSP power supply at a good wattage for a good price, and it doesn't say 80+ or anything on the sticker on the side, just copy-paste the model # in to google and add pdf onto the end and see if you can pull up it's specs. Might be better than it appears. 😀 Typically most FSP power supplies are at least solid enough to provide full power at their rated spec. I've owned a couple before.

386SX wrote:

Tetrium wrote:

386SX wrote:

Interesting! I never built anything that high level so I always choosed low middle end psu. Actually I've a cheap Corsair VS650 for newer builds and a Enermax EG365 old type for retro builds. The Enermax is great inside with many good components but I don't know about its efficiency.

You learn something new every day! 😀

I don't think it matters much if your rig is under 200W or so when measuring from the wall. Even my Phenom II never seemed to consume that much, but that's also because I stick to mid-range graphics (and I'm not using Seasonic or another absolutely top-brand PSU, I've always gone for the next best thing like FSP) and I'm more of a "playable framerates" guy and don't even have any HD displays here. I actually tend to disable all the extra stuff like smoke because to me that's just an extra annoyance I don't want to have to deal with 😜

I usually buy too the cheap version of a good quality psu. What do you think about FSP psu?

Same here. But I do do my homework before I buy a brand new one. When it comes to second hand, I'll read up if the PSU happens to have any significant design flaws or other quirks I should know about. And the very first thing I do with second hand PSUs is to open them up and have a look inside (and clean it up if it looks ok).

In the old days to me FSP was the easiest PSU to find that wasn't total crap (everyone knew Antec was good, so these were never found for cheaps) and that nobody knew about (everyone wanted Antec and Seasonic so FSP units were kinda forgotten and thus as cheap as the really crappy ones) and the older units (like for instance 300W units with 30A 5v ratings) were pretty good for things like Athlon XP (one of my FSP 300W units powered a Barton for years, though only with a Radeon 9600).

I do noticed that they seemed to start cutting corners a bit after they had already started making 350W PSUs (but mostly with stuff that isn't really super critical), but usually they should never blow up in your face and I'm still never worried whenever I use one (after having taken a look inside that is 😜).

350W FSP powering Q6600 and 8800GTS? Yes it can!
500W FSP powering an AM3 Athlon II quad core + crossfire HD6850? Yes it can!

But I wouldn't really recommend using such configurations for a prolonged period.

I've also had good experiences using other brands that base their units off of FSP designs, like the older OCZ PSUs (actually used 3 of those, each one for a couple years at the least and in different builds) or older AOpen PSUs. I've never had one burn up on me, even though I ended up using FSPs in literally way over half of all my builds (I'm talking about >>20 builds here). The only PSU that gave me problems was some Premier that was lighter than air. Replaced that "300W" sucker with an ancient 235W PSU that knew how to get the job done and all problems went flying away! 😁

I have used many other brands though, but for most of the other brands I ended up using only a single unit (one Antec, one BeQuiet, one Tagan, one Coolermaster, etc) and imo one cannot judge an entire brand that's existed for years by judging a single product.

So what tools and tests do people use to load the system?

Especially for Windows 98 there might be less options.

Prime 95 for the CPU is a good program I like to use. What about graphics cards?

Tetrium wrote:

(I'm talking about >>20 builds here). The only PSU that gave me problems was some Premier that was lighter than air. Replaced that "300W" sucker with an ancient 235W PSU that knew how to get the job done and all problems went flying away! 😁

I have used many other brands though, but for most of the other brands I ended up using only a single unit (one Antec, one BeQuiet, one Tagan, one Coolermaster, etc) and imo one cannot judge an entire brand that's existed for years by judging a single product.

Lot of builds there! I usually build many config on the same case... 😁
After having changed a many psu I'll never buy again those nobrand or unknown brands lighter than air psu... I've seen some that would not even power a low freq Duron (edit: Duron 1200 and 4,8V and lower)... and on the sticker like 40-50A on the 5V rail... even the plug cable couldn't stay well inserted on it...
Another one I had in the past... first run at boot... in the bios 17V instead of 12V... 7V instead of 5V... something like that... oh yeah! vga and other components gone. How was technically possible I'll never know..I could expect lower values.. but higher..

PhilsComputerLab wrote:

So what tools and tests do people use to load the system?

Especially for Windows 98 there might be less options.

Prime 95 for the CPU is a good program I like to use. What about graphics cards?

On Win98 I would use 3dmark2000 and 2001!

386SX wrote:

Lot of builds there! I usually build many config on the same case... :D After having changed a many psu I'll never buy again tho […]
Show full quote

Tetrium wrote:

(I'm talking about >>20 builds here). The only PSU that gave me problems was some Premier that was lighter than air. Replaced that "300W" sucker with an ancient 235W PSU that knew how to get the job done and all problems went flying away! 😁

I have used many other brands though, but for most of the other brands I ended up using only a single unit (one Antec, one BeQuiet, one Tagan, one Coolermaster, etc) and imo one cannot judge an entire brand that's existed for years by judging a single product.

Lot of builds there! I usually build many config on the same case... 😁
After having changed a many psu I'll never buy again those nobrand or unknown brands lighter than air psu... I've seen some that would not even charge a phone battery... and on the sticker like 40-50A on the 5V rail... even the plug cable couldn't stay well inserted on it...
Another one I had in the past... first run at boot... in the bios 17V instead of 12V... 7V instead of 5V... something like that... oh yeah! vga and other components gone. How was technically possible I'll never know..I could expect lower values.. but higher..

I don't need to empty an old build as I have plenty spare ones (most are kinda crappy though)

Yup, that's definitely a "crap"-PSU 🤣, but 17V and 7V?..that's insane! 😲

I'd prefer a 300W FSP over a cheap 600W "Happy Smile" PSU any day! (The brand "Happy Smile" is a fictional name and it better stay that way 🤣)

This is why I actually use some (expendable) test hardware whenever I try out anything I recently gotten. I actually developed a workflow and used separate boxes for stuff that still needed to be visually inspected and cleaned, then tested and sticker with date and "works!" and only then would I add these parts to my spare parts boxes. Heck, I actually used a box of all parts that failed one test (yes I stickered these as well with an "X" plus date) as most would actually work in another test setup or after having fixed something that I had previously missed.

In the end I ended up frying very few parts for all the stuff I did, even though I have no degree in computers or in software whatsoever! It's just a matter of common sense and learning I guess.

386SX wrote:

Hi,
lately having this plug eletric power meter always on (why oh why did I buy it.. 😁) I see that all the config I build use quiet a lot of eletric power. My Barton 3200+ config always end up consuming around 105>115W at idle no to mention on cpu load where it can easily go up to 130W and more.

I became really disappointed in AMD Socket A after finding out about the power usage. Power saving features are almost non-existant/malfunctioning, except on the very last chipsets. More info.

Some old readings of mine:
Sandy Bridge Pentium G620 on Gigabyte Z68MX, Cheap FSP PSU, Intel HD Graphics, Laptop HDD:
Idle Desktop Windows XP = 42 Watt / 3D game running = 60 Watt.

AMD Sempron 3000+ on Asus A7V600, PSU Aopen 350W, HD3450 Graphics, Desktop HDD, Santa Cruz Sound:
Idle Desktop Windows XP = 76 Watt (Edit: That is with working power saving) / 3D game running = 110-125 Watt.

I also tested 7 different Power Supplies back then:
Mini-ITX mainboard with Intel Core Duo T5600, HD5450 Graphics.
System Turned off: 2 to 8 Watt
System waiting in boot menu: 57 to 66 Watt
System idle Windows XP desktop: 41 to 50 Watt
System running 3D game: 67 to 76 Watt.
Strangely the upper Watt scores were attributed to my Chieftec 360W. Which I actually bought new a little earlier, and it was not a budget model at all. Okay it was overpowered for that system but 8 watt when off does not help. The Best scores were from a Powerman 235W and an FSP 250W.

gerwin wrote:

I became really disappointed in AMD Socket A after finding out about the power usage. Power saving features are almost non-exist […]
Show full quote

386SX wrote:

Hi,
lately having this plug eletric power meter always on (why oh why did I buy it.. 😁) I see that all the config I build use quiet a lot of eletric power. My Barton 3200+ config always end up consuming around 105>115W at idle no to mention on cpu load where it can easily go up to 130W and more.

I became really disappointed in AMD Socket A after finding out about the power usage. Power saving features are almost non-existant/malfunctioning, except on the very last chipsets. More info.

Some old readings of mine:
Sandy Bridge Pentium G620 on Gigabyte Z68MX, Cheap FSP PSU, Intel HD Graphics, Laptop HDD:
Idle Desktop Windows XP = 42 Watt / 3D game running = 60 Watt.

AMD Sempron 3000+ on Asus A7V600, PSU Aopen 350W, HD3450 Graphics, Desktop HDD, Santa Cruz Sound:
Idle Desktop Windows XP = 76 Watt / 3D game running = 110-125 Watt.

I also tested 7 different Power Supplies back then:
Mini-ITX mainboard with Intel Core Duo T5600, HD5450 Graphics.
System Turned off: 2 to 8 Watt
System waiting in boot menu: 57 to 66 Watt
System idle Windows XP desktop: 41 to 50 Watt
System running 3D game: 67 to 76 Watt.
Strangely the upper Watt scores were attributed to my Chieftec 360W. Which I actually bought new a little earlier, and it was not a budget model at all. Okay it was overpowered for that system but 8 watt when off does not help. The Best scores were from a Powerman 235W and an FSP 250W.

Interesting.
Strange thing of the Barton was that the Powernow feature was there in the XP-M model and it scaled the frequency quiet well just like the K6-2+. Even older xp-m models with the older core had them.

Tetrium wrote:

I don't need to empty an old build as I have plenty spare ones (most are kinda crappy though) […]
Show full quote

386SX wrote:

Lot of builds there! I usually build many config on the same case... :D After having changed a many psu I'll never buy again tho […]
Show full quote

Tetrium wrote:

(I'm talking about >>20 builds here). The only PSU that gave me problems was some Premier that was lighter than air. Replaced that "300W" sucker with an ancient 235W PSU that knew how to get the job done and all problems went flying away! 😁

I have used many other brands though, but for most of the other brands I ended up using only a single unit (one Antec, one BeQuiet, one Tagan, one Coolermaster, etc) and imo one cannot judge an entire brand that's existed for years by judging a single product.

Lot of builds there! I usually build many config on the same case... 😁
After having changed a many psu I'll never buy again those nobrand or unknown brands lighter than air psu... I've seen some that would not even charge a phone battery... and on the sticker like 40-50A on the 5V rail... even the plug cable couldn't stay well inserted on it...
Another one I had in the past... first run at boot... in the bios 17V instead of 12V... 7V instead of 5V... something like that... oh yeah! vga and other components gone. How was technically possible I'll never know..I could expect lower values.. but higher..

I don't need to empty an old build as I have plenty spare ones (most are kinda crappy though)

Yup, that's definitely a "crap"-PSU 🤣, but 17V and 7V?..that's insane! 😲

I'd prefer a 300W FSP over a cheap 600W "Happy Smile" PSU any day! (The brand "Happy Smile" is a fictional name and it better stay that way 🤣)

This is why I actually use some (expendable) test hardware whenever I try out anything I recently gotten. I actually developed a workflow and used separate boxes for stuff that still needed to be visually inspected and cleaned, then tested and sticker with date and "works!" and only then would I add these parts to my spare parts boxes. Heck, I actually used a box of all parts that failed one test (yes I stickered these as well with an "X" plus date) as most would actually work in another test setup or after having fixed something that I had previously missed.

In the end I ended up frying very few parts for all the stuff I did, even though I have no degree in computers or in software whatsoever! It's just a matter of common sense and learning I guess.

Of the ones I have the Enermax I use for XP builds, is the best in term of components. Big heatsinks, a lot of protections circuits, good capacitors... general nice build and it's really old.
The VS650 has not many components but still better than others.

386SX wrote:

Interesting.
Strange thing of the Barton was that the Powernow feature was there in the XP-M model and it scaled the frequency quiet well just like the K6-2+. Even older xp-m models with the older core had them.

I did not mean Powernow downclocking, although that would help. The problem is that in >95% of the socket A systems the processor is always kinda running 100%, even when doing nothing. It cannot rest when idle. STPGNT and HLT instructions are not enabled by default. Just like how an idle MS-DOS prompt can make a modern CPU sweat.

My main rig gets around 80watts idle and 190watts under full load (Unigine Valley).

Will edit to add a few DOS machines in a bit.

486DX2/66 with floppy, optical drive, VLB graphics, ISA controller and sound, and 1GB Seagate Medalist: 49 watts no matter if its idle or running Quake timedemo.

Pentium Overdrive 200MMX with floppy, optical, PCI TNT2 M64, ISA sound, and 20GB Quantum Fireball lct: 48 watts idle, 50 watts during Quake timedemo.

K6-2 550 with floppy, optical PCI TNT2 M64, ISA sound, and 20GB Quantum Fireball lct: 59 watts idle, 60 watts running Quake timedemo.

Good info 😀

I will for sure try to incorporate power readings into future videos, I can see how this could be of interest.

1x Pentium III Tualatin 1.4 @ 1.57GHz on MS6095 Slocket
Tekram P6BX-A(n)
768MB (3x 256) of PC-133 SDRAM
1x Nvidia GeForce 6200
1x Sound Blaster AWE64 Gold
1x Promise FastTrack S150 TX2 PCI card (to provide a SATA header for the SSD)
1x USB 2.0 PCI card
1x Ethernet PCI card
1x SSD
1x HDD
1X DVD
1x 3.5" Floppy
2x 120mm + 1x 92mm + 1x 80mm fans
250W 'Oh Deer' PSU (of about 70% efficiency tops)

· PC turned off but plugged to wall (live +5VSB): 3.4 W
· Max peak observed while booting up: 98 W AC from the wall (about 70 W DC)
· Idling, HDD on/off: 81 / 75 W AC from the wall (about 57 / 53 W DC)
· Grinding Prime95 large FFTs, HDD on/off: 93 / 87 W AC from the wall (about 65 / 61 W DC)

= = = = = = = =

1x Pentium 4 Preshott 3.4HT @ 3.9GHz
ASUS P4P800
3GB (2x 512MB + 2x 1GB) of DDR400
1x Radeon 9800 Pro
1x SSD
1x HDD
1x DVD
1x 3.5" Floppy
3x 120mm + 2x 80mm fans
Enermax ENP500AGT PSU (83% max efficiency)

· PC turned off but plugged to wall (live +5VSB): 1.2 W
· Max peak observed while booting up: 250 W AC from the wall (about 207 W DC)
· Idling: 152 W AC from the wall (about 126 W DC)
· Grinding Prime95 large FFTs: 278 W AC from the wall (about 230 W DC)
· AIDA64 Stability Test (CPU-FPU-Cache-Memory-GPU): 265 W AC from the wall (about 220 W DC)
· Logged Off: 153 W AC from the wall (about 127 W DC)
· STR3 Suspend to RAM: 3.5 W

TELVM wrote:

250W 'Oh Deer' PSU (of about 70% efficiency tops)

Got a good fire insurance policy?

^ Heh heh worry do not, this particular specimen has been thoroughly modified with oversized all-jap caps, enhanced cooling and other improvements, and in fact works like a charm considering its age.

386SX wrote:

By the way... Sempron dual 1,45Ghz, Radeon R5 with on demand power load: 22W in idle oh yeah. 😁

I take it you're running an AM1 APU rig? How's the performance?

mr_bigmouth_502 wrote:

386SX wrote:

By the way... Sempron dual 1,45Ghz, Radeon R5 with on demand power load: 22W in idle oh yeah. 😁

I take it you're running an AM1 APU rig? How's the performance?

I only tried web on it and it's good imho if we consider its power consumption.
Maybe with the internal Radeon R3 igp it would have lower watt too. But it's the lower end of the am1 line so low clock and 1333 limited memory. But for a bit more speed I would like to try the only a bit more expensive Athlon AM1 that has also 2MB cache.