I see a W942508DH!

[largon]

Interesting...

While downloading BH-5's datasheet from Winbond's archive, I wondered what would Google find about "W942508DH"...

To my surprise, it actually gave a result:
http://www.google.com/search?client=...en&q=W942508DH

Altavista can find another online IC dealer listing them.

This is seems to be a new generation of the same die as for example CH-5. If we get lucky, Winbond might one of these days begin pushing these things to module mfgs.

Too bad the chip is atm rated only 7.5ns. I wonder has Winbond even tested what DH could really do...
I guess their engineers really aren't so keen on OC'ing.

[craig588]

Remember when CH first came out? DH will be just like it.

[Vapor]

Remember when CH first came out? DH will be just like it.

For those of us who don't remember when CH first came out....care to explain?

[largon]

Yes, most of the time CH-5 couldn't do tRCD 2. Only the last samples that reached DIMMs, could do it.

We can hope though.

[Vapor]

Hmmmm, good to know, thanks.

It's amazing how well the new CH is doing then (read: VX, Redline, etc....most UTT), has much changed since Winbond made it?

[largon]

We don't actually know what happened to CH-die, but it's speculated that Infineon helped Winbond to optimize the die or the process.
CH-UTT was the result.

[celemine1Gig]

IMHO, it wasn't the CH-chips that were "bad" at te beginning. They were performing the same as they do today. The difference was that anyone expected it to best BH-chips, which did 200MHz 2-2-2-x at unter 2.8V. Well, CH-chips couldn't best it, they couldn't even do 200Mhz at 2-2-2-x, but "only" at 2-3-2-x. Although, it was still good, the dilemma was that noone tested it at above 3V at that time. That's why nobody wanted to have CH-RAM at the beginning and its reputation was kinda bad. But that's only my opinion.

And about DH-chips:

They are available for some time already and some people wrote about having harddrives which had a DH-chip as cache-RAM installed. But I doubt that we'll ever see a DDR-RAM stick with DH-chips, as DDR2 is making its way, thus great investments in DDR-RAM would not be that intelligent, I think.

[$ilver]

Jo, DH is not that new .. ive seen it last week on some of our SCSI Harddrives in our IBM Netvista Server (36GB HDD ~3Years old) ..

@Cele , i tested some Kingston Hyper X 3500 mit CH5 2years ago with 3,35v , all they did was 245MHZ 5-2-3-2... My 5month old UTT-CH did at that voltage 255 5-2-2-2.. so i dont think its the same chip

[high5]

IMHO, it wasn't the CH-chips that were "bad" at te beginning. They were performing the same as they do today. The difference was that anyone expected it to best BH-chips, which did 200MHz 2-2-2-x at unter 2.8V. Well, CH-chips couldn't best it, they couldn't even do 200Mhz at 2-2-2-x, but "only" at 2-3-2-x. Although, it was still good, the dilemma was that noone tested it at above 3V at that time. That's why nobody wanted to have CH-RAM at the beginning and its reputation was kinda bad. But that's only my opinion.

mine also.

And about DH-chips:

They are available for some time already and some people wrote about having harddrives which had a DH-chip as cache-RAM installed. But I doubt that we'll ever see a DDR-RAM stick with DH-chips, as DDR2 is making its way, thus great investments in DDR-RAM would not be that intelligent, I think.

i was just about to post the same. i've personally seen many dh winbond's on various hard drives, vgas etc. i even think my friend still has some ancient vga that utilizes dh75 ic's. not sure whats the entire p/n though...

[celemine1Gig]

Jo, DH is not that new .. ive seen it last week on some of our SCSI Harddrives in our IBM Netvista Server (36GB HDD ~3Years old) ..

@Cele , i tested some Kingston Hyper X 3500 mit CH5 2years ago with 3,35v , all they did was 245MHZ 5-2-3-2... My 5month old UTT-CH did at that voltage 255 5-2-2-2.. so i dont think its the same chip

Remember, the boards have to play nice with the sticks, too. At the time, CH-RAM was released, most boards couldn't even use the full potential of the BH-RAMs. I didn't mean to say that the new CH-UTT chips are exactly the same as the old CH-chips. I meant to say that if you test old CH-RAM today, they will most likely perform as good as new CH-UTTs.

[largon]

Those HD caches are W986416DH - 8MB chips.
This is W942508DH 32MBs, just like BH-5, CH-5 and UTTs.

They are totally different chips.

[largon]

It has been an assumption that the first letter of for example "BH-5" means 175nm, really it isn't.
W942508AH- is 175nm too, as is the the BH-flavor.

My guess is that it's just a chip generation prefix just like "F" in Samsung's F-TCCD.

Is there a chip decoder somewhere in Winbond site?

[high5]

doesn't neccesarily matter. as celemine mentioned, dh's were always used as cache memory and probably will stay that way. the 7.5ns rating of particular dh's you mentioned also suggests that.

[largon]

Give me a single pic of W942508DH as a cache.

btw. Winbond binned 7.5ns W942508BHs too.

[high5]

give me a single pic of any dh as system memory.

winbond rated bh dies @7.5ns years ago, at the time when 133mhz ram was the state of the art tech. that dh is _presently_ not rated faster than 7.5ns and you expect it to develop into some high-end ram?!? keep it real.

[craig588]

IMHO, it wasn't the CH-chips that were "bad" at te beginning. They were performing the same as they do today. The difference was that anyone expected it to best BH-chips, which did 200MHz 2-2-2-x at unter 2.8V. Well, CH-chips couldn't best it, they couldn't even do 200Mhz at 2-2-2-x, but "only" at 2-3-2-x. Although, it was still good, the dilemma was that noone tested it at above 3V at that time. That's why nobody wanted to have CH-RAM at the beginning and its reputation was kinda bad. But that's only my opinion.

Maybe you didn't, but I was doing the Vdimm=3.3 line mod for a long time, before CH5.

[largon]

32MB chip is DEFINATELY NOT a HD cache chip.
As you said High5, "keep it real"

It may be, that Winbond doesn't have enough fab to produce this for a full scale use in pc DRAM (yet?). That's one good reason for the lack of higher binnings.

I have not seen any referring to this chip anywhere before.
AFAIK nobody has reported seeing this chip on anything.
It must be a new chip.

[high5]

aha... okay.

[celemine1Gig]

Maybe you didn't, but I was doing the Vdimm=3.3 line mod for a long time, before CH5.

Exceptions prove the rule.

[largon]

That proverbial HD cache is W986416DH.


8 megabytes, 175nm, 3,3Volts - and it's plain SDRAM.

Edit: what an earth are these datasheets doing in http://www.ortodoxism.ro??

[DaWaN]

Winbond DH is problably the UTT that is fixed onto VX and AA4T TwinMOS mem. As those are totally different chips as CH-5. I don't if it's DH but that type of UTT isn't defently NOT the same as 2 year old CH-5

[largon]

DaWaN,
There are representatives of Mushkin, TwinMos, G.Skill and OCZ that all have stated that their UTT-RAM uses only CH- and BH-UTT.

[DaWaN]

DaWaN,
There are representatives of Mushkin, TwinMos, G.Skill and OCZ that all have stated that their UTT-RAM uses only CH- and BH-UTT.

That's true but as I said, keep in mind that it's possible UTT because those are really different as CH-5, just compare a UTT to a CH-5, UTT can do 2-2-2-5 and much higher clockspeeds

[largon]

There are numerous reports about CH-5 doing 2-2-2-x 250+.


3 possible explanation:

1: Winbond had problems with the C generation of the IC. Before CH-5 was discontinued they managed to fix (most of) them. -> CH-UTT is a resurrected fixed CH-5.
*I prefer this theory*

2: CH-5 = CH-UTT

3: Winbond had some collaboration with Infineon. Result was a fixed CH-5: CH-UTT.
*I like this too*

[STEvil]

or they just got better at producing ch-x after a while (like how some weeks of bh-x suck).