Tonight I make a change of RAM, in the PC Office, I briefly remember MB ABIT IP35 with C2D E4600 8 GB DDR2 800 CL6, on Windows 7 I didn't get a better score going from 6 to 8 GB, it's quite strange (in my opinion), so I want to try to change all the RAM, with others from 1 GB DDR2 800 CL4, from here you could understand, if the switch from CL6 to CL4, improves the performance of the RAM, in any case I would say of Yes, unfortunately those modules are 1 GB, the two I didn't find them, these If they don't show better performance, I think I'll reuse them on other PCs like this one.

I will also see to do some benches, for comparison with those made a few days ago, we can only hope that the difference is quite substantial, otherwise I will have lost time ⏱, in the sense that the advantage of the CL4 could be lost with the greater amount of total RAM

Updates on the PC Office, I performed the change of RAM, well do not expect miracles, there is an improvement but it is practically imperceptible, however since I use Windows 7 X86, and that it exploits just over 3 GB, now with 4 it remains unused about 3/4 of 1 GB.

Momentarily, I leave it like this, aesthetically I'm not bad at all, and the whole is getting interesting, maybe it would look good in case 🏘 with windowed side panel, but of these I don't have any, the best I can host it is a Centurion, I still have to do ️ some additions, and then I realized that the service pack 1 is missing, this update can improve video performance, in the past on Internet PCs it has risen by about half a point.

I made a few benches, but I didn't have the time ⏱ to make a comparison, the difference, as I anticipated, if there is is minimal.

I have grouped all the latest CPUs I have purchased lately, the cost has been really low, including shipping, it goes from just under 3 € to 7.00 €, some of these (a couple) I already have them, they would be duplicates, but they are useful for any motherboards that do not support faster CPUs.

It goes from Core2 Duo 775, to Core2 Quad 775, I have never had the Core2 Quad, while of the Duo I have several, one of the Quads I could put it in the PC Office PC, the alternative is a Core2 Duo E8500, which I keep aside (in collection), before deciding whether to replace the Core2 Duo E4600 currently installed, I would like to have an idea if the fastest one with two cores is better, or one of the two with four cores, then a separate speech deserves overclocking, it seems that starting from the lowest frequencies, you can make the FSB jump quite easily, example 800 to 1066 or 1333.

Finally among the other CPUs I have an i7 1156, the motherboard is of course ASUS (amone my favorites ), I have to fix at least one pin, it won't be easy but I'll try to straighten it.

Returning to the Willy 1700, I performed some tests, first I took the 80-pin cable from the MSI MS-7312 or K9MM-V, thanks to this the error reported at startup in the BIOS, has disappeared, it remains that of the floppy not connected, which I could delete it, but I need it to stop the startup screen, and be able to take the necessary shots.

Trying other readers, the one currently used does not need an 80 pin cable, while another does, unfortunately it does not read the disks, for this reason as written before, I will have to try the ASUS 754 with Sempron, maybe this time I can install a Windows.

In the Willy I followed the performance index, to see if the change of the cable can get a better score, actually there doesn't seem to be a difference, this makes me think that the unit is slow of its own, and that something faster needs to be used.

I've uploaded some programs, the same ones I have on the PC Office, we'll see as soon as I'm done with the benches, the results, I hope they're not too below average, however finished these, I'll see to proceed to an upgrade, I'll switch to a slightly faster CPU, always Willy, maybe I have a 2000, if I don't find it I have a 1900, there won't be much difference but it can go, unfortunately there are no Willy from more than 2000 MHz!!!

Regarding the Windows 7 bench running with the Willy 1700, I noticed that the value of Desktop Aero has risen from 3.5 to 3.7, little I know, but I don't think it can be attributable to the 80 cable, maybe it could depend on a few more MHz, however it's fine, better to have earned them than lost.

I have among the various loaded programs, BMX, I had to run it as if the OS was 98-ME, because otherwise it won't boot, the results are visible in the screens, later I'll do other benches.

PC@LIVE wrote on 2024-03-01, 18:47:

I have grouped all the latest CPUs I have purchased lately, the cost has been really low, including shipping, it goes from just under 3 € to 7.00 €, some of these (a couple) I already have them, they would be duplicates, but they are useful for any motherboards that do not support faster CPUs.

It goes from Core2 Duo 775, to Core2 Quad 775, I have never had the Core2 Quad, while of the Duo I have several, one of the Quads I could put it in the PC Office PC, the alternative is a Core2 Duo E8500, which I keep aside (in collection), before deciding whether to replace the Core2 Duo E4600 currently installed, I would like to have an idea if the fastest one with two cores is better, or one of the two with four cores, then a separate speech deserves overclocking, it seems that starting from the lowest frequencies, you can make the FSB jump quite easily, example 800 to 1066 or 1333.

I was researching dual to quad performance issues a while back and came to the conclusion that for about 2005 to 2015 windows software, aside from multithreaded encoding or work like that, just desktop apps and games, because of poor utilisation of the multiple cores in run of the mill stuff, a quad needed to get within 20% of a dual core clockspeed to be faster than it. i.e. 2.5Ghz quad is better than 3Ghz dual core, but 3.2Ghz dual core is getting away from the quad again. This may not be the case on the 2015 to date software, where things really use as many cores as you can throw at it. Early windows XP it's even about the same for dual vs single, but late XP dual was king. I think there are +/- 5% variations with cache size.

Anyway, with that, if you are purposing them retro prior to 2008ish then it might be fastest core clock you can get whatever number of cores is more important, so even the celeron 4x0 might be a consideration, especially if you can get it up to insane 4ghz plus clocks, the pentium dual variants are also easy to push to high clocks, since cache makes heat so a bit less goes further up. Then 2005-2015 it's whether you can get a C2D clock at least 20% higher than your quad clocks, then 2015 to modern, you want both highest instruction set support and highest all cores clock total, so 45nm and 3ghz plus quad preferable. I don't know a good point of compromise between older 65nm quad vs 45nm dual, I guess if you can't run something needing sse4.1 you can't run it, but it's not going to be stellar on a dual unless you've got really high clocks.

BitWrangler wrote on 2024-03-02, 17:41:

PC@LIVE wrote on 2024-03-01, 18:47:

I have grouped all the latest CPUs I have purchased lately, the cost has been really low, including shipping, it goes from just under 3 € to 7.00 €, some of these (a couple) I already have them, they would be duplicates, but they are useful for any motherboards that do not support faster CPUs.

It goes from Core2 Duo 775, to Core2 Quad 775, I have never had the Core2 Quad, while of the Duo I have several, one of the Quads I could put it in the PC Office PC, the alternative is a Core2 Duo E8500, which I keep aside (in collection), before deciding whether to replace the Core2 Duo E4600 currently installed, I would like to have an idea if the fastest one with two cores is better, or one of the two with four cores, then a separate speech deserves overclocking, it seems that starting from the lowest frequencies, you can make the FSB jump quite easily, example 800 to 1066 or 1333.

I was researching dual to quad performance issues a while back and came to the conclusion that for about 2005 to 2015 windows software, aside from multithreaded encoding or work like that, just desktop apps and games, because of poor utilisation of the multiple cores in run of the mill stuff, a quad needed to get within 20% of a dual core clockspeed to be faster than it. i.e. 2.5Ghz quad is better than 3Ghz dual core, but 3.2Ghz dual core is getting away from the quad again. This may not be the case on the 2015 to date software, where things really use as many cores as you can throw at it. Early windows XP it's even about the same for dual vs single, but late XP dual was king. I think there are +/- 5% variations with cache size.

Anyway, with that, if you are purposing them retro prior to 2008ish then it might be fastest core clock you can get whatever number of cores is more important, so even the celeron 4x0 might be a consideration, especially if you can get it up to insane 4ghz plus clocks, the pentium dual variants are also easy to push to high clocks, since cache makes heat so a bit less goes further up. Then 2005-2015 it's whether you can get a C2D clock at least 20% higher than your quad clocks, then 2015 to modern, you want both highest instruction set support and highest all cores clock total, so 45nm and 3ghz plus quad preferable. I don't know a good point of compromise between older 65nm quad vs 45nm dual, I guess if you can't run something needing sse4.1 you can't run it, but it's not going to be stellar on a dual unless you've got really high clocks.

Thank you very much for the precious suggestions, so according to another user's suggestion, the simplest thing to do would be to use (in the absence of a 9550 or 9650) the Q6600, but not at standard frequency, but raising the FSB from 1066 to 1333, perhaps it will be necessary to cool the CPU better, and I don't know whether to overvoltage a bit, the final result would be 3.00 GHz and 8MB of cache, a little less than the 12MB of the 9550-9650, but still more than those with 4 -6MB, in this regard I saw that the user Phil, in his YouTube channel, has a Q6600 at 3.00 GHz, but if it was not possible to go up to 3.00 GHz, I would try and I think it is possible, go up to FSB 1200, which corresponds to 2.70 GHz, certainly easier to reach, by the way the motherboard should be an ABIT IP35, but I have other MB 775s, and in those I can possibly put other Qxxxxs, I will be able to say more about this later.
As you may have seen I am currently working on a Willy 1700, if I wanted I could change it with a 2.40/400/512, but I want a PC 478 P4 Willy, but having other MB 478s, I also have the 533 and 800 ones, I even have a 533@667, on an ASUS P4B533 2.67@3.33 NW, and then other Prescotts.

It seems "weirdly easy" to hit 3Ghz with most of conroe+ class CPU, seem to start hitting resistance where "bad" ones fall out around 3.2, but even then some go to 3.6ghz on stock volts. Then with massive attention to cooling, voltage tuning and keeping the motherboard happy, they can go another gigahertz on top of that which is crazy for anything before or since.

BitWrangler wrote on 2024-03-02, 19:33:

It seems "weirdly easy" to hit 3Ghz with most of conroe+ class CPU, seem to start hitting resistance where "bad" ones fall out around 3.2, but even then some go to 3.6ghz on stock volts. Then with massive attention to cooling, voltage tuning and keeping the motherboard happy, they can go another gigahertz on top of that which is crazy for anything before or since.

Yes, thank you very much, this confirms that it is not difficult to go up to 3.00 GHz, and for me they are sufficient, going beyond well will also be possible, but I don't think it makes a big difference, among the 775 PCs, I have an E4700 2.6@3.0 GHz, everything at standard Vcore, the biggest jump at standard VCore is the P4 2.67@3.33 that I wrote about before, if I had a 2.8 I could try 3.5 GHz, which however is not possible with Core2 CPU, for the highest FSB and lowest multi .

A little while ago I made a couple of benches on the Willy 1700, I don't know how useful they can be, but they can be compared with other similar PCs, I would think for example of an Athlon XP 1700+, well it is not obvious that the fastest one is one of the two, probably the value of a single bench does not give the idea if the PC is better or worse, you need to make different benches, with HW as similar as possible, to note that an XP 1700+ actually goes slower than the 1700, so already for this reason, if it approaches the P4 1700+ it is a nice result.

I managed to load one of those bench programs, which I had in other PCs, but in the Willy and others I could not load, it is the 3DBench, I tried unsuccessfully a previous version, then with the 2001 SE, it started and completed the installation.

To sum up it's about this PC:

MB ECS P4IBAS (478 with 3 SDRAMs)

CPU P4 Willamette 478 1700/256/400

VGA AGP Gigabyte GV-R955256T 256 MB

3X 512 MB PC133 Elpida chip

HD WD Caviar 307AA 30.7GB

This in summary of the system, I will make some changes later, but currently it is so.

The score obtained seems to me quite below average, but I could be wrong, maybe it's not so?!?

Yeah it's seeming about half what that CPU should do with that card, or half what that card should do with that CPU. R 9550 should be somewhere near 9600SE .. this is more like Radeon 7000VE, or 9550 on a PIII 900.

BitWrangler wrote on 2024-03-03, 16:09:

Yeah it's seeming about half what that CPU should do with that card, or half what that card should do with that CPU. R 9550 should be somewhere near 9600SE .. this is more like Radeon 7000VE, or 9550 on a PIII 900.

Thank you very much, I suspected it, it must be said that it could be due to the RAM, in that MB ECS there are three SDRAM banks, the normal PC100-133 so to speak, I have another PC with a similar MB, and it has never shone in speed , despite being a 2800/512/400, instead a 423 with 1 GB RDRAM and P4 2.00/256/400, is quite good, I have to see if I can start both and make a comparison, even if they use different VGAs.
Just out of curiosity, I loaded the same program on the Office PC, an ABIT IP35 with C2D E4600 and GF GT620 2 GB DDR3, I'm just under 20000, here too I don't think it's in the average scores?!?

Since I had a few minutes I uploaded a few other programs, one of bench and other diagnostics.

The bench one would be AquaMark3, the final score can be seen (in the image), I think it's low, both for the not very fast CPU, and for the memory, also not very fast.

Looking at the Astra32 diagnostic, I noticed that the audio driver is missing, unfortunately I have no speakers on the bench, but I can connect them to try the audio, I will see to recover the suitable driver, that is, assuming that there is an audio driver for Windows 7 X86.

I saw that some bench have problems, the 3D Mark2000, I can't install it, honestly I don't know the reason, maybe in the official website I could find a working one??!!??

Other than that, today I tried in the PC Office some of the programs, which I had used yesterday in the Willy, then saw the most recent hardware, I loaded the 3D Mark2003, and I booted the benches, ditto for AquaMark3 and HWinfo32.

The results I don't know if they can be average, or below? Anyway I think I'll do more tests soon, and then I think I'll start mounting both PCs, in their respective cases 🏘, although I have to check if you can find the audio drivers of the Willy PC, obviously for Windows 7 X86, for the previous versions I'm pretty sure they are easily found.

The 3D Mark2000, I managed to install it, downloaded the file from the Phils website, and installed it, first in the Office PC and then in the Willy PC.

Commenting on the results, it seems useless to me, if those of the 3D Mark 2001 SE, were below average, here it is not that I do better, however it must be said that the PCs are not optimized, both with regard to the drivers, and with regard to the BIOS, let's say that there may be the possibility, to do better by doing some settings, however I was interested in installing this program, unfortunately I have not yet found ️ the 3D Mark 99, also I have to try the 3D Mark 7, I have to install it and see the scores.

About Windows 7, I saw that there is a second service pack, it's not actually called a service pack, I don't know if it's a simple collection of updates, or it's a package of updates, which avoids loading them individually, I know very little about it, and if that SP2 is available in various languages

Next motherboards awaiting repair and testing, a 486SX with soldered UMC U5SX CPU, has damage from battery acid leakage, and I noticed that a couple of SMDs are missing in the battery area, I don't know exactly what they are, but maybe you can see in photos of identical cards, if you can figure out what they are.
The other is an ATX Siemens D969, with Intel HX chipset, Cirrus Logic integrated video card, with expandable memory, by adding two memory chips.

PC@LIVE wrote on 2024-03-06, 18:22:

Next motherboards awaiting repair and testing, a 486SX with soldered UMC U5SX CPU, has damage from battery acid leakage, and I noticed that a couple of SMDs are missing in the battery area, I don't know exactly what they are, but maybe you can see in photos of identical cards, if you can figure out what they are.

Hope you get it going, those UMC are fun to run, there's a quite similar board here https://theretroweb.com/motherboards/s/a-trend-atc-1411b not sure if you can see what is missing clearly enough.

BitWrangler wrote on 2024-03-06, 19:09:

PC@LIVE wrote on 2024-03-06, 18:22:

Next motherboards awaiting repair and testing, a 486SX with soldered UMC U5SX CPU, has damage from battery acid leakage, and I noticed that a couple of SMDs are missing in the battery area, I don't know exactly what they are, but maybe you can see in photos of identical cards, if you can figure out what they are.

Hope you get it going, those UMC are fun to run, there's a quite similar board here https://theretroweb.com/motherboards/s/a-trend-atc-1411b not sure if you can see what is missing clearly enough.

Thank you very much friend, I will check the images as soon as possible, hoping to be able to see the missing SMDs, one of them should be a resistor, and it will not be easy to understand the value or be able to read it, I will take photos of the detail, I hope that after a good cleaning, not other damage emerges, in addition to the visible ones, I like the 486, I have some to restore, but I like this one because it has the UMC CPU, looking better, the 72 pin RAM slot needs to be changed, because it has a broken hook, then I think you need the cache and TAG chips.