Crew Strunkenbold Posted July 22, 2020 Crew Posted July 22, 2020 (edited) The problem is that those 4 metal dots on chips side are very common. Also Infineon BT and PowerChip have those metal dots in this arrangement but first one is usually not rated at CL3-4-4 and the later are quite rare chips and afaik nobody did OC them. Hynix however made several revisions of its chips, so OC behavior could improve over time because of better manufacturing process. If the GEIL were late produced, maybe you are just lucky. It's actually not quite correct to speak of "D43" as its just the speed bin. The actual chip depends on chip density and revision. Anyway this is most likely a Hynix 512Mbit late Revision D4 IC. D4=CL3-4-4 DDR400. I think when the party with DDR ram was over, because of the appearance of DDR2, interesting ICs appeared on the market but little is known about their OC potential. Even though I dont think there is still a chip out there which would be able to beat TCCD or BH to be OC relevant. But its an interesting field nonetheless as new DDR1 ICs were engineered till 2010. For example the successor of Infineon CE, the DE IC was available in 2008. Also Qimonda DF BGA ICs were introduced in 2009 and I saw results from someone with CF-5 already reaching DDR-600... You can find basically nothing about ELPIDA or Powerchip but who knows if a late revision of them is something interesting for us. Also Elixir is something like Kingston, they don't have their own chip design or even a fab. They just buy untested chips and place them on a pcb. Most Elixir chips I saw looks like to be Infineon BT. ProMos and Mosel are IMO also Infineon BT. Edit: Later ProMos and Mosel chips should be Elpida as Infineon left the Joint Venture and Elpida jumped in. Edited July 23, 2020 by Strunkenbold 4 Quote
Tzk Posted July 23, 2020 Posted July 23, 2020 (edited) Hynix got at least 3 different revisions (BT, CT and dt), two densities (256mbit and 512mbit), lead free packages (for example DTP-D43). I tested 256mbit BT-D43 and DT-D43 in nforce 2, all sticks were 256mb singlerank. The BT-D43 were able to pass 32M 3-4-4 at 257mhz (chipset limit) and the DT-D43 did 250 2.5-3-3 but failed above 255mhz, regardless of voltage or Timings. All sticks were genuine hynix so no binning, just four random 256mb sticks. All four Sticks passed 200mhz 2.5-3-2, cant remember If cl2 was possible. Trcd 2 however didnt post. I've yet to try better hynix sticks like OCZ on NF2. Afaik OCZ made DDR500+ Hynix sticks. All in all Hynix seems to run well on NF2 and does achieve similar results to TCCDs. However TCCD is a nightmare on NF2 at 250MHz+... Edited July 23, 2020 by Tzk Quote
Crew Strunkenbold Posted July 23, 2020 Crew Posted July 23, 2020 Someone made a nice summary of DDR1 at reddit: https://www.reddit.com/r/overclocking/wiki/ram/ddr1#wiki_hynix Seems like there were up to rev G for 256Mbit and rev E for 512Mbit. However I've never seen such high revisions. 1 Quote
Tzk Posted July 23, 2020 Posted July 23, 2020 Yep, i've only seen up to D Revision sticks in the wild. I still want to try some Hynix D5's on NF2 or at least higher binned sticks. Maybe 2.5-3-3 is possible at 260MHz+ if the board is capable. However even Infineon made D Revision Chips (DE-5) which are quite unknown. Same as Samsung ZCCC (BGA) or Qimonda CF-5 (also BGA). Quote
Crew Strunkenbold Posted July 23, 2020 Crew Posted July 23, 2020 1 hour ago, Tzk said: However even Infineon made D Revision Chips (DE-5) which are quite unknown. Same as Samsung ZCCC (BGA) or Qimonda CF-5 (also BGA). They are rare but actually you can still find them today on ebay. AFAIR my DE-5 stick was better than all my CE-5s but nothing too phenomenal. Also I've never tested the single Qimonda DF-5 I own, can't even tell if it is working but also never saw one again, so these are actually ultra rare. I also guess those revisions, at least for Infineon, means a die shrink. Samsungs also had different revisions. Most, if not all TCCD are rev F. from 2004. But honestly dont know much about them. ZCCC can also be at least Rev F. However, I dont know how they overclock. Quote
Tzk Posted July 23, 2020 Posted July 23, 2020 I got Rev. F TCCDs with production date of week 42 2005, so there were newer chips made. However my Geil Ultra-X (unknown date, probably 2004) are by far better. The Corsair 3200XL v1.2s can't even reach their spec on Socket A (250MHz 2.3-3-3) but fail to pass 240Mhz 3-3-3. That's disappointing für TCCDs. The Ultra-X at least reach 250Mhz 2.5-3-3 primestable. I guess the NF2 just isn't doing well with these Samsung chips. Same for 1gb Infineon CE-6 sticks and 512mb singlerank. Those fail to pass 210Mhz on my NF2 boards, regardless of voltage and timings. However 256mb Infineon BT-5 ( Corsair 3200C2PT v4.x) reach 250MHz 3-3-3-8 without issues. The Nforce is indeed quite picky when it comes to certain ram chips. Winbond works incedibly well, Hynix too. Haven't tried Micron or some other type of Infineon yet. Quote
Sparks.nl Posted July 23, 2020 Posted July 23, 2020 (edited) I have got some golden dragons which do 260 2-2-2-5 2t on nf2. And 1 XL kit makes it up to 260-265 c2.5 boot as well. Those Geil refuse 1t on nf2 and nf4 Edited July 23, 2020 by Sparks.nl Quote
digitalbath Posted July 23, 2020 Posted July 23, 2020 (edited) My Qimonda DE-5 sticks do a good job on my nF2 systems, but I assume they do a better job on nF4 systems. They only needed ~2,48V for this 264MHz / 3-3-2-8 / 32M run. nF2 is limiting here. CPU-Z is wrong with the RAM size. Edited July 23, 2020 by digitalbath 2 Quote
Crew Strunkenbold Posted July 23, 2020 Crew Posted July 23, 2020 2 hours ago, Sparks.nl said: I have got some golden dragons which do 260 2-2-2-5 2t on nf2. And 1 XL kit makes it up to 260-265 c2.5 boot as well. Nice kit you've got there. CL2 @ 260Mhz ? Thats crazy. Quote
Sparks.nl Posted July 24, 2020 Posted July 24, 2020 (edited) 8 hours ago, Strunkenbold said: Nice kit you've got there. CL2 @ 260Mhz ? Thats crazy. It is just that 1 seems broken. So that is why I never look at them. I never tested pi though. Edited July 24, 2020 by Sparks.nl Quote
Tzk Posted July 24, 2020 Posted July 24, 2020 (edited) 12 hours ago, Sparks.nl said: Those Geil refuse 1t on nf2 and nf4 I've run Geil Ultra-X (2x512mb TCCDs) at 250MHz 3-3-3 1T on NF2 through 32M The A7N8X doesn't seem to like CL2.5 with Merlin Romsips... I've had 2.5-3-3 250MHz 24/7 stable on DFI Infinity years ago. It's achievable for sure Booting them above that isn't an issue, but getting it stable enoughfor 32M is... TCCDs are very picky and the NF2 just doesn't like them at all. Edited July 24, 2020 by Tzk 1 Quote
Sparks.nl Posted July 24, 2020 Posted July 24, 2020 (edited) From what I saw the mem controller in the nb chip is very picky. One can handle something, others can not. But there are kits who seem to do better overall. I noticed this week that my Epox board likes ds 256MB memory more then ss. My Abit boards like ss more. Edited July 24, 2020 by Sparks.nl Quote
TASOS Posted July 24, 2020 Posted July 24, 2020 Anyone care to mod an Epox bios ... (or anyone already have a modded one) ? It's about the EP-8RDA3I (non Pro) ... sample photo It's the board without the Sata ports ... and has the older 256kb bios. The latest version available is the rda34729.bin (29/07/2004) rda34729.bin Quote
Tzk Posted July 24, 2020 Posted July 24, 2020 13 hours ago, Sparks.nl said: From what I saw the mem controller in the nb chip is very picky. Yes. I got a A7N8X-E Deluxe which won't run 210Mhz fsb without voltmods. The silicon quality seems to fluctuate a lot... With mod it can't reach 220Mhz. Must be the worst -E Deluxe ever made. Quote
Sparks.nl Posted July 24, 2020 Posted July 24, 2020 3 minutes ago, Tzk said: Yes. I got a A7N8X-E Deluxe which won't run 210Mhz fsb without voltmods. The silicon quality seems to fluctuate a lot... With mod it can't reach 220Mhz. Must be the worst -E Deluxe ever made. That is probably the worst I have “seen”. I reccomend testing all Winbond dimms you have and maybe tccd as well. In my experience there could be 1 or 2 dimms that run great and surprise you. Quote
digitalbath Posted July 26, 2020 Posted July 26, 2020 On 7/24/2020 at 9:29 PM, TASOS said: Anyone care to mod an Epox bios ... (or anyone already have a modded one) ? It's about the EP-8RDA3I (non Pro) ... sample photo It's the board without the Sata ports ... and has the older 256kb bios. The latest version available is the rda34729.bin (29/07/2004) rda34729.bin 256 kB · 1 download I modded your BIOS for the EPOX 8RDA3I. Before flashing, please make sure, that CMOS register 73-78 are free. You can check it with this program. It should look something like this (in this screen CMOS register 73-78 are used by my mod): Some things left to say: the FSB Frequency is limited in the BIOS by 250MHz. 256KB BIOS! BPL is 3.09 by default, I left it in the original version. I do not own good BPL Versions for 256kb BIOSes. This BIOS versions should work, but they are completely untested! I will not be responsible for any damages! Having a programmed backup bios chip before flashing is not a bad idea. 8RDA3I_EBED_A6.zip 8RDA3I_ED_A6.zip 8RDA3I_X1_A6.zip 3 Quote
TASOS Posted July 26, 2020 Posted July 26, 2020 6 hours ago, digitalbath said: I modded your BIOS for the EPOX 8RDA3I. Before flashing, please make sure, that CMOS register 73-78 are free. You can check it with this program. It should look something like this (in this screen CMOS register 73-78 are used by my mod): Some things left to say: the FSB Frequency is limited in the BIOS by 250MHz. 256KB BIOS! BPL is 3.09 by default, I left it in the original version. I do not own good BPL Versions for 256kb BIOSes. This BIOS versions should work, but they are completely untested! I will not be responsible for any damages! Having a programmed backup bios chip before flashing is not a bad idea. 8RDA3I_EBED_A6.zip 385.63 kB · 1 download 8RDA3I_ED_A6.zip 385.73 kB · 1 download 8RDA3I_X1_A6.zip 385.56 kB · 1 download Thank you I will test your bios next weekend. Yes i already have a pre-programmed backup bios chip for this board. 1 Quote
Sparks.nl Posted July 31, 2020 Posted July 31, 2020 (edited) Ok so we talk about alternatives. And yes, dfi is expensive when found... So I bought this defective board for €80... and it is working again (without Firewire ??). Edited August 1, 2020 by Sparks.nl 1 Quote
Mr.Scott Posted July 31, 2020 Posted July 31, 2020 (edited) Expensive. Needs a re-cap badly. Pretty much all my Ultra B is good for is high FSB and SPI. I use other boards for everything else. Edited July 31, 2020 by Mr.Scott Quote
Sparks.nl Posted August 1, 2020 Posted August 1, 2020 (edited) 9 hours ago, Mr.Scott said: Expensive. Needs a re-cap badly. Pretty much all my Ultra B is good for is high FSB and SPI. I use other boards for everything else. Well since last time I had seen one being sold was at least a year ago and that one did €180-200. So €80 for defective is expensive, for working board is Ok. I am hoping so that it does a nice job in fsb speeds. Oh and I recap myself. I had caps allready. I would not turn this on before. Edited August 1, 2020 by Sparks.nl 1 Quote
GRIFF Posted August 1, 2020 Posted August 1, 2020 Dfi Ultra-B is the great board. Don't be surprised at the price, it's a fairly rare card. 1 2 Quote
Thorn Posted August 1, 2020 Posted August 1, 2020 it is.. feed it it with some TCCD. I`m curious :) Quote
Sparks.nl Posted August 1, 2020 Posted August 1, 2020 6 hours ago, GRIFF said: Dfi Ultra-B is the great board. Don't be surprised at the price, it's a fairly rare card. If you ever find one... 1 Quote
Thorn Posted August 1, 2020 Posted August 1, 2020 (edited) ever heard of the "LIST" ? https://www.hardwareluxx.de/community/threads/amd-sockel-a.584473/page-29#post-27073135 we got some working ones in germany. Feel free to contribute Edited August 1, 2020 by Thorn Quote
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.