Quote:
Originally Posted by QerryTerry
He works for twinmos so i dont think he bought it any where.
Printable View
Quote:
Originally Posted by QerryTerry
He works for twinmos so i dont think he bought it any where.
Great news. Hope they are performing on par with the good old BH-5 at least! :D
@ Rabbi_NZ , i remember ppl saying that 512mb sticks of BH-5 didn't oc as well as 256mb. i dunno the actual reason why, but it was an known issue.
only some few 512mb sticks managed to oc as the 256mb... the thing is, i bet we won't see this happening with these new BH-5 UTT. ;)
PS: i noticed now... Corsair pc3500... they had to do well at 245mhz @ 3.2v mate. :p:
remember this TwinMOS are rated pc3200 with 5ns chips.
let's wait for more testing from Alonso! :toast:
DDR400 is -5ns, just like DDR500 is -4ns if i understand correctly...Quote:
Originally Posted by enok
but im crossing my fingers about 2x512mb as well, im hoping that the new PCB makes it overclock better with 512mb sticks, memory controllers have also came a long way since that time.
I would love to throw some of this new BH5 at my 3.0E! Sadly I've got to wait for something like Hipro5's Maximizer before I can even think about better RAM. :/
i can proof that... corsair had built the best bh-5 dimms on the market. Mine have done 245@ 2-2-2-11-1t on nf7-s 2.0 with dirty punk... but now my corsair is going to r.i.p. it doesnt like high voltages any more, and it brings up errors in memtest with stock speed/timings. BUT! I've ordered TwinMOS SP UTT :) i will give infos about max clock.... I will ship my old corsair for rma...
EDIT:
for all who understand german :)
http://www.stock-world.de/detail/new...age=1&a=623100
I would have to disagree and say OCZ PC3500 Platinum Limited Edition is the best BH-5 ever made, BH-5 on a BrainPower PCB :slobber: Its impossible to find but I saw 1GB of it sell for $400 almost instantly. I'm still searching for some, PM if you're willing to sell some!
BP pcb gave no extra to CH-5 UTT.. it would be nice to see it with BH-5 UTT.
every module that was ever made using BH5 had 5ns chips ;)Quote:
Originally Posted by enok
I have to diagree, you obviously never tried to make the damn stuff work on an Nforce2 Ultra 400 mobo :mad:Quote:
Originally Posted by HiJon89
I've only seen a couple people with 1GB PC3500 Platinum LE kits but from what I've seen they've all reached 250FSB 2-2-2-5 with 3.3V or less, something no other manufacturers' BH-5 could do consistently.Quote:
Originally Posted by GazC
But with A64 I assume... no nforce2 board, no via kt333/400/600 board could use their "potential"... why that? I would not consider that the best memory made. Remember the hyperx 3200 bh-5 I have, those sticks are great on p4, a64 and nf2 etc.Quote:
Originally Posted by HiJon89
Go figure.
Yea, KHX BH5 beats Corsair BH5 on average in terms of oc. The KHX PCB was custom for BH5 chips and it showed.
OCZ and Mushkin L2 Black are at the top of the heap in terms of best BH5, but OCZ PC3500 Platinum LE takes the crown. :)
KHX PCB is standard JDEC PCD :D ;)
Nope, modified JEDEC. :rolleyes:
EDIT: The chips on KHX even have more leads going to them then the BH5 on other sticks. I have some KHX 3000 BH5 in front of me right now.
it's a known CHIPSET issue... 2x512 BH5 does jus fine on a good mobo/cpu comboQuote:
Originally Posted by enok
The only reason 2x512 oc's worse than 2x256 is because there are twice as many chips being overclocked. So your chances are less, plus it puts more stress on the pcb. However, this isn't a huge deal, on average your oc's will be 5-10mhz less at most I would think. And Yes, it is a chipset limitation, people say this from the nforce 2 days when it didn't like 2x512, and also same situation with 2x512 an 1t timings.
NForce2 does have a wall with 512M "double sided" sticks. Further the BH5's had a 2T limitation ....ie:
The Twinmos UTT 2x 512M are running just fine 1T, get same max FSB/Voltage whether set to 1T or 2T actually. But bandwidth results are higher with 1T, of course. Wall at 245Mhz, regardless of CPC.
The BH5 2x 512Mhowever wont run 1T at the same high FSB, 2T it runs fine and hits the exact same wall. In my case, 245Mhz.
2x 256M is a whole different issue. I dont have 256M sticks of BH5 or UTT but my 2x 256M "old" CH5 does 245Mhz and each stick alone breaks 250Mhz. People have gotten HUGE FSB results from 256M sticks of BH5...Ive seen as high as 280-290Mhz on NForce2.
So there is definitely something in the physically in the chipset or in unknown settings that put a severe limitation on 512M stick results....even though the EXACT same chips/settings on 256M stick hit 10-20-30Mhz higher.
Whther this is due strictly to the extra 16 memory chips loaidng the bus beyond the ability of the NF2 northbridge to drive, or there is some setting left un-altered by any known bios/tweaker that NEEDS to be changed to make double-banked operation happy.... beats me.
Ex: what if bank-interleave was going on, set by default in all the bios's of NF2 market, and noone ever knew it or ever altered the setting. notice how bank-interleaving setting in the new DFI-SLI bios has a drastic effect.... makes you think, obviously the NF2 can handle high FSB.... ask Macci, LastViking, etc.
It all goes back to NVidia's secrecy, even on a "dead" product. Release the damn data manual on the chipset and we will very quickly get to the bottom of what the ABSOLUTE MAXIMUM is out of the NForce2. As it is, on things like the NF7-S, I cant even LOOSEN the Alpha timings in the bios, so I have to BOOT slower than I should, tweak the Alpha's with WPCRSET and then up the speed in Clockgen. And that is a group of settings we KNOW about.
All a moot point, just buy a damn A64 board and forget about NF2 once and for all...at least thats my plan. And now I have 2Gigs of Identical UTT that will get me to my HAPPY PLACE :banana:
EDIT: Forgot to note, single stick of BH5 actually does 252Mhz max in the NF7-S, but I have to raise FSB after WPCRSET loosens the Alpha's to 5,4,6,3,3,4,5
from the tight default coded in the Abit bios. TicTac looked into fixing the Alpha's but it proved too intricate a change. I look forward to TicTac's results on 256M sticks of UTT in Infinity board.
uwackme, try to lauch a team of special forces to get you the data sheets ... ;)
:CTF:
ya but the point i was highlighting was, Corsair was rated pc3500 and not pc3200. :p:Quote:
Originally Posted by Rabbi_NZ
excuse me... but wth is brainpower pcb and what is the difference to "standard" pcb's ?
mfg
i tried to get some today and went at least into 10 shops but couldn'd find any!
bh5 got a bad rap for not doing 1t with 1gb kits because of the nf2 memory controller. once more people start dusting off their sticks and making them scream in a nf4 board we'll see some (more) awesome results.
khx did worse than corsair or mushkin, in my experience. i had pc3000, pc3200, and pc3500 bh5 in my dfi infinity, and corsair and mushkin hit 5-15 mhz higher every time. I've just started playing with nf2 again (got a lanpartyB) and I've made my sticks do 34s superpi, 245mhz 2-2-2-11 cpc on. That's 3.3v. Cpc off they do 260+ easily, but that's where the board maxes out. Once I get some longer bolts I can mount a tec on the nb and see if that helps.
as much as i'm dying to upgrade to dfi nf4 I can't afford mobo, cpu, and video card all at once. if dfi releases a 939 pin nf3 mobo I'll be all over it. so, unfortuanly i can't go proving how well 1gb kit will do in better platform :(
anywho, looking forward to this new bh-5, but if it doesn't show improvement over old stuff might have to wait for D :)
mine are standard JEDEC. whats different about yours then?Quote:
Originally Posted by EnJoY
i thought BH5 had 66pins iirc. how can it have more? thats interesting. could you maybe post some pics so i can comapre them to mines?
im in ... :DQuote:
Originally Posted by SAE
Hehe... who else? :D
Hey guys while we are on this topic I think you'd like to see this thread on Ocforums I made:
http://www.ocforums.com/showthread.p...43#post3476943
____Quote:
Originally Posted by Sentential
Point is it ISNT bh5. Its a new version rather a pure unbinned and unmarked BH which everyone calls UTT. Sure it is *technically* BH5 in every way, but it is being made side by side Infineon products in the same fab. Twinmos has a healthy supply via Winbond so the odds of you getting UTT in a pack of premium Twinmos is really damn high. Id go for it :banana4:
I'd also like to add that this "UTT" is apparentally being made from different wafers and are a differnt process size from any of the former winbond products. BH5 and all former Winbond tech is still extinct. This is something wholy new.
Also just for reference this stuff is *tehcnically* although not directally, manufactured by Infineon. <in the same plant>.
I also want to clarify that the pure infineon based BH isnt capable of decent timings/speeds at all. From what I heard 4-4-4 @ 200mhz is a best case scenerio. The same applies to the Mosel/ProMOS stuff. Neither ones take volts well and often die instantally when givein 3.2v+
The only stuff capable of doing 2-2-2 is the one using 100% winbond tech. These raw BH chips are available from people like Twinmos and possibly Adata.
However Twinmos has *the* highest concentrations due to the "stragetegic alliance" ;)
Alonso,
Do you have any idea when the "distribution channels" will be getting this stuff into UK stores?
Sentential,
cheers for the update mate.
BUT, you keep saying UTT/BH... whereas from every other source I have read it seems UTT=CH die, not BH die.
CH=0.13u
UTT=0.13u
UTT are blank CH dies.
Jus to clearify:
The "new BH" TwinMOS is getting will be on the BH die I think... according to Alonso.
I'm just saying previous ones seem to have been on CH die...
Well the point is that they are BH esk hence why I said that. Its just how you word it but the concept is still the same :toast: However they are NOT former winbond tech so technically they arent Ch5 either they are something completely different.Quote:
Originally Posted by Rabbi_NZ
From my understanding of things this "UTT" is basically Infineon fabing using Winbond's specs hence why they are :banana::banana::banana::banana:ty/need so much voltage :stick:
- Winbond BH5 = 0.175µm
- Winbond CH5 = 0.13µm
- Infineon/UTT BH5 = 0.15µm
- Infineon/UTT CH5 = 0.13µm
Are there any known differences other than size of the dies in these four chips?
They use a differnt type of wafer and size. They use standard Infineon wafers afaik. However the dies are similar, probabaly not exactally the same. I'd have to ask specifically about that.Quote:
Originally Posted by largon
Are you sure about this?Quote:
Originally Posted by largon
If this is true these are exciting times... but I havent heard anything like this before.
What are your sources?
I cant say who my source is but I will say it is from a leading dram manufacturer. So I can confirm for 100% that the UTT is by proxy Infineon made. Its made in their fabs and that is enough to call it Infineon in my book.Quote:
Originally Posted by Rabbi_NZ
The designs are still winbond but its being outsourced and then sold to Twinmos.
I thought old bh5 (winbond bh5) was 15nm?
Im not sure what process the UTT is made on...but cha know what. I can ask tonight :cool: This person isnt on atm but I will send an email. :toast:Quote:
Originally Posted by kryptobs2000
I was refering to the fab process Sentential, can you confirm new and old BH5 use a different process?Quote:
Originally Posted by Sentential
PS, a VW Beetle made in a Chev factory will still be a VW Beetle in my books... unless of course they change the makeup of the vehicle
Yes they do use a differnt process. I dont know which but I can confirm for 100% that both:Quote:
Originally Posted by Rabbi_NZ
A) different die process
B) different size die
C) differnt size wafer its cut from
What's the closest thing that Infineon makes thats comparable to this?
If you wanna know right now look at Mosel, it uses UTT based chips that would be your best guess. However I will ask in an email because frankly I really didnt get into specifics however I will tonight, IF this person knows it.
Ill aslo see if I can get a product spec.
well in that case, any word on dh5?
Let us know bro :toast:Quote:
Originally Posted by Sentential
Yea Ill do that. Ill see what kinda dirt I can dig up. However I wanna warn you, there may not be a product ID.Quote:
Originally Posted by Rabbi_NZ
This person said that Infineon/Winbond sold the bare ICs and didnt even bother to bin them, or label just litererly blank ICs. Thus leaving/forcing OCZ and Twinmos to manually do so :bananal:<and yes im serious> go thru and bin them themselves. Thus why they are so friggin expensive and so rare.
So its entirely possibel that it wont have a name <or at the least lacking an ns rating>
At least we'll know how to find the TwinMOS BH/UTT and CH/UTT... those AA4T and 1A4T codes are all we need :)
im kinda comfused after this. What Ram do I want to buy to get BH-5? and BH-5 is better than CH-5 as of now? Thansk
Just spoke with my "source" tonight :toast: This person was going to be talking to the Winbond techs anyway tonight and is gonna ask those questions along with providing serial IDs and tags :banana4:
Twinmos PC2700. These sticks are on the egg and they are labeled AA4T, they got the mtec chips... Hmm, wut IS twinmos up to?
EDIT: 37$ shipped, if truly BH5 or CH5, this is awesome and I'm pickin up 2 sticks for my 754 rig (vnf3-250 & 2800+) thats comin in a couple more days.
DO NOT GET THOSE!!!Quote:
Originally Posted by stevemedes
Those are the UTTs I was warning about. Those are the pure Infineon UTTs, they are not Winbond's
Sent, can u point me in the right direction of which to get? I'm thinkin of goin with either the 256meg or 512meg sticks of twinmos SP... they seem to be out of stock though.
If you want the best shot it would be any of the PC3200. The pure infineon garbage wont do PC3200 speeds with any sort of tight timing.Quote:
Originally Posted by stevemedes
from what im told 4-4-4 is typical for Infineon based stuff, ie the PC2700
So even though this stick of 512 has the MTEC chips, theres still a decent chance its UTT? just wanna 2x check before I order. I'm a noob to the A64 platform. Not used to being so picky about RAM lol.
Its possible that is UTT, however THIS:
http://www.newegg.com/app/ViewProduc...218-062&depa=1
IS utt. Hell just look at the reviews ;)
Quote:
Originally Posted by Sentential
lol
Is it? Haven't bothered, my 2nd and 3rd rig is using this memory due to cheap cost :P Gotta take some time to play with it later. :)
:doh:I had meant that it is possible that the PC2700 is real UTT, however the PC3200 GB pack is.Quote:
Originally Posted by esdee
I bit the bullet and ordered the 1gb kit that Sent recommended. I doubt I'll regret it. Thanks for the help.
When I get my system all set up I'll post my results.
Not a problem. In a couple hours / tommorow Ill have those serials you guys have been chomping at the bits to get.Quote:
Originally Posted by stevemedes
I dont think you'll regret it. thats what I got and it works great. HAHAHa that one of those reviews on newegg are mine.
AFAIK, the Chaintech VNF3-250 mobo im getting only allows up to like 2.8vdimm. These UTT chips like much higher voltage than this correct? If true, can someone point me to a Vdimm mod that works (or someone they know who did it and it worked.) I have relatively limited soldering skills. All i've really done was make my own baybus. Hopefully this should be enuff soldering knowledge to get this mod done, cause I wanna o/c the hell outta my system. :banana:
so these UTT chips are good for DFI NF4?
http://forums.techpowerup.com/showthread.php?t=129Quote:
Originally Posted by stevemedes
Im gonna be doing that in the next few weekends, whenever I have time to my VNF3-250
;) ;)Quote:
Originally Posted by Magnj
UTT works well in just about any AMD board....which remids me I outta ask this person what the deal is with the compatability issues on Intel.
I havn't heard of any bh5 (or utt) problems with intel, only tccd problems.
They exist for sure.Quote:
Originally Posted by kryptobs2000
Some boards have the timing options to get UTT working, but those that dont have enough tweaking options simply wont play
Friendly word of warning on the memory from newegg... 4 sticks I have doesn't behave well at all... It definitely is not for low timing etc. I am wondering if there were different revisions of this memory, etc. I believe I got these 4 sticks about a month ago or so. (same part# etc though).
Still cheap memory for my 2nd and 3rd rig :)
Ive actually seen and heard the opposite. If anything TCCD does perfectally fine in Intel. Granted it doesnt clock as high for chipset reasons but hey it gets teh job done.Quote:
Originally Posted by kryptobs2000
Just ordered 30GB´s of this 1A4T stuff from TwinMos europe (via distributor) :D
ETA less than two weeks.
OMFG :owned:Quote:
Originally Posted by The Stilt
Damn you gonna swimm in memories :D
Nope, we have sold over 150GB´s of TwinMos UTT sticks in less than two months, so I don´t have enough time to swimm in them before they´re gone :DQuote:
Originally Posted by VoRtAn_MaDgE
**Waiting for "reseller of the year" award to arrive from TwinMos**
'Wink' Alonso 'Wink' ;)
I´m dying to see if the new process (0.175µ -> 0.15µ) brings any increase in overclockability of these new BH-5 chips.
They are building em on a new process now? Hmmm maybe I should get a set.Quote:
Originally Posted by The Stilt
i wanna buy my stuff from the egg but all i hear are bad reviews about the overclocking-bility of theirs products ( CPU's and now RAM) whats a good retailer Monarch?
the lol was about the reviews ;)Quote:
Originally Posted by Sentential
@the_stilt
damn the_stilt :D ... what happend to all those UTT anyway!?!?
Sold them :DQuote:
Originally Posted by esdee
Alonso told it the first time on page #1 of this thread...Quote:
Originally Posted by Sentential
in lower voltage too ;) the sweet spot for the UTTs proccess seemed to be 3.4VQuote:
Originally Posted by The Stilt
Is Twinmos (and OCZ) the only mfg that applies UTT's?
Interesting... Ill ask around and see what I can dig up.Quote:
Originally Posted by esdee
:toast:Quote:
Originally Posted by Sentential
Atleast A-Data and Diamond also use this chips.Quote:
Originally Posted by largon
Yea the person I was talking to hasnt contacted me about what their discussion with Winbond was. Perhaps this was it. Kinda sux when there is a time difference. i was too tired to stay upQuote:
Originally Posted by esdee
:dammit:
Wouldnt suprise me seein how Adata is OCZ's cheif supplier ;)Quote:
Originally Posted by The Stilt
@ Stilt..... FFFFKKKKK... I cant afford Euro's, Im a poor American FFFFKKKK :slobber:
you aint poor, otherwise you wouldnt have bought 4 sticks of UTT :D
and you have X800XT...geeze.... :stick:
Little update:
Finding out more information concerning UTT. UTT is infact its actual name. Lemme try to clarify the issue:
Infineon makes the UTT modules. To avoid paying tariffs on the modules which is about 15% Twinmos/OCZ / etc request that these modules are left unmarked.
These unbinned modules can and are used in a wide variety of Infineon/Mosel products but most people dont know them because they are normally binned and given a label.
With UTT they are blank and leave the reseller to manually bin the RAM. Thus the ICs are either hit or miss. Frankly im a little confused as to how this is legal but apparentally according to what this perons says it happens all the time and is "buisness as usual"
The "wonder" of UTT is nothing more than heavily binned and volted Infineon.
Btw it is moving to a smaller process very soon. Ill have an update on that later
So.....
Winbond gives Infineon their whizbang recipies
Infineon creates the same cake, but in their ovens... maybe making the cake smaller to reduce costs and increase profits
Infineon sells these cakes to all bakeries at a reduced price, but with no guarantee they will be up to any desert standards
The bakeries now tastetest these cakes and label them accordingly so Joe-CakeLover-Average doesn't get a chocolate cake when he was really after carrot cake
Quality of cake testings is OCZ > TwinMOS > Mtec > Infineon
Am I close Sentential? hehehe
i think u hit bullseye, Rabbi_NZ! :p:
smaller process?... hummm... still on DDR? DRR2... maybe DDR3?
Quote:
Originally Posted by Rabbi_NZ
rabbi you confused ( :D ) me but i know i like one thing ... Black Forest ;)
Ill take a Double choc & butter cream cake pls
I hope these mems arrive to the stores soon... :slobber:
Some results would be nice too.
I don't buy it.
Winbond makes the ICs FOR Infineon, Infineon does not make the ICs. I believe that it was Alonso, the TwinMOS rep who stated that Winbond is under an agreement that does not allow it to sell Winbond branded chips, not that they can't produce chips. See attached news release. It states that the Winbond UTT chips are not labeled. It also states that Infineon only used part of the capacity, which is why the excess capacity is being utilized.
http://www.globalsources.com/gsol/I/...0000055550.htm
Infineon ICs are physically different than Winbond made chips. See attached for comparisons if you need to. Most here KNOW what Winbond chips look like.
http://www.biwa.ne.jp/~yok/TENJIKAN-DDR.htm
Winbond uses Infineon technologies (R&D) because Winbond could not afford to do this on their own.
I have seen nothing posted that would lead me to believe that Infineon is producing UTT, however, they may have some UTT based ram, since they are buying UTT from Winbond. We have analyzed this UTT to death in 90 pages over at the TwinMOS thread. NOTHING has changed...it is Winbond.
Yes he did. I got a little more clarification today. Here's the *real* issue at hand.Quote:
Originally Posted by enok
Any forign semi-conductor that is bought in the USA is taxed like a mother:banana::banana::banana::banana::banana::ban ana:.
What happens is that Infineon produces the base ICs for Winbond. Winbond then rebadges/packages them in what small fabs they have. They tweek em and get em going.
Then to avoid tariffs and taxations they do not bin them or lable them their tru form "Infineon".
If its Infineon it immediatly gets a 15% tariff, for Hynix its as high as 57%
Winbond sells the bare ICs to their good buds at Twinmos and pocket the 15% for themselves since the bare ICs are non-taxable. Overall its pretty shady but according to my sources its just "buisness as usual"
Btw there is no name for this stuff other than its pure code name "UTT". It is neither Infineon OR winbond because it isnt binned by them.
THink of it this way. All winbond is doing is selling repackaged Infineon chips by the lot. Twinmos buys them and bins them accordingly. Some do very well most dont.
Ive asked for what micron they are using and expect that to come soon. But yes Infineon's fabs are going to a smaller die process, Ill find out what as well..
However the REALLY interesting fact is this.... what were those base ICs to be used in Infineon RAM? Since they are UTT based technically we should start seeing Infineon sticks that do UTT timings and voltage. Perhaps not as good but it should be a clear sign
Also from the same source I have *BIGGER* news that you all would like to hear....
Getting word via the grapevine that Samsung is in the testing phase of the successor to TCCD <no im not kidding> Details are still sketchy and they do not know if they will go public with it or not.
Depending on the consumer demand we may or may not see TCCE <guessing>
Either way its binned clocks are expected to come in at a whopping DDR550 making it the first and only binned RAM clocked that high <tccd is 500>
did you mean 550, because you said whopping 500, highest ever binned, and then tccd is 500?
nvm, you fixed it. incredible, binned at the fab at 550. that means that the sellers could probably bin that stuff at like 650 for their expensive performance ram.
dude, you really got me confused now. Infineon sell winbond there own chips?
Is infineon based over seas and winbond based in the us?
I edited my post just after you replied :owned:Quote:
Originally Posted by Skip
[Edited post for clarity] Gimme a break :( spent the last 6 hrs at work and had 5 classes today. Sen = So tired :stick:Quote:
Originally Posted by Shroomalistic
Infineon makes the bare / base ICs and winbond beefs them up and repackeges them. Thats how UTT is made.
Producer = Infineon
Repackager = Winbond
From what I understand this is all done in the same factory since Infineon aquired almost all of Winbond's dram assets and then split em with Mosel
Its no differnt than buying reved up CTD50 from Adata. Ya'll know what i mean
yea :)
how long was tccd on the market? if they released tcce within the next 6 months, would it be on the market about as long as tccd was before ddr2 comes out?