Need help with developing ram =)

[ChongL]

Hey guys...I'm not sure if this belongs in this forum but I think it's the closest to the topic at hand. I just got hired by a local ram company who wants to enter the gaming/overclocking market to overclock and beat the crap out of their ram. And I'm supposed to give them feedback.

Now I know it won't be BH-5 or TCCD, but just wondering what I should tell them in terms of a balance of timings vs. raw speed.

Any input from you guys would be appreciated since you guys are all extreme here. Thanks again!

[Major]

2-2-2-5 > 250mhz @ 3.2v

2.5-3-3-7 > 300mhz @ 2.8v

[P_1]

Hey guys...I'm not sure if this belongs in this forum but I think it's the closest to the topic at hand. I just got hired by a local ram company who wants to enter the gaming/overclocking market to overclock and beat the crap out of their ram. And I'm supposed to give them feedback.

Now I know it won't be BH-5 or TCCD, but just wondering what I should tell them in terms of a balance of timings vs. raw speed.

Any input from you guys would be appreciated since you guys are all extreme here. Thanks again!

what?? are they making their own ram chips? If so then ask them to develop ram that will scale high with low voltage and low timings(2-2-2 or 1-1-1 ) like 275mhz with 2-2-2 timings at 2.7v would be very nice

[Magnj]

1-1-1-5 @350MHZ please.
No realy though. There should be a happy medium. Tight timings around 250-260 is great. What kind of chips?

[shadowing]

I'd like to ask what company?

It'd be nice to have anything that's 1-1-1-5, or 2-2-2-5 at 250, or more MHZ.

[ChongL]

I start tomorrow so I'll see how things are run first. But their test room has an FX-53 for me to play with . I get paid to overclock now!!

I'll try and tell them to balance low latencies (2-2-2) like you guys said with around 250-260 MHz. Hopefully I'll have some influence and it will end up helping us all. The company is gigaram btw...they sell some stuff on newegg right now

Thanks for the responses once again!

[[XC] moddolicous]

see if they can make mem that runs 1-1-1. I dont think there is any memory out right now that can do that.

[P_1]

I start tomorrow so I'll see how things are run first. But their test room has an FX-53 for me to play with . I get paid to overclock now!!

I'll try and tell them to balance low latencies (2-2-2) like you guys said with around 250-260 MHz. Hopefully I'll have some influence and it will end up helping us all. The company is gigaram btw...they sell some stuff on newegg right now

Thanks for the responses once again!

so what are they doing? are they making new ramchips or are they just making ram like corsair and buying chips and pcbs and putting them togetehr and selling

[ChongL]

i think they do both. They started off producing just ram chips and selling them but now they actually make the chips and the stick itself I think.

[Sephious]

I'd like to see a low latency (2-2-2-5) memory that scales well, even at very high voltages, such as above 3.6V.

[One_Hertz]

Well there are really two main different ram markets TCCD and BH-5. These are quite different one relies on timings another on raw speed and for some processors TCCD will be better, and others BH-5. I really don't think that selling something in between will help because if you are trying to reach a high HTT with a low multi you will just grab yourself some nice tccd chips and overclock. That is good mostly for amd processors because you can not really run 300+ htt 24/7 on an intel platform cause that would force you to a 50% oc, which most chips can not handle, so bh-5 tends to be more popular with intel. As far as I know both the chip types are about equally used on amd platforms. If you made something in between it will not be bought much because it will no longer have a specific use and will not be better than either type. You should just come up with your own TCCD and BH-5 and try to beat the current chips by raw speed or timings, than you will definately get some sales.

[jiff]

simple really, look at the interest when someone manages to get 270+ 2.2.2.5 from a new product we all . Then order some and when we only get 260. If a company speed binned a product that could do minimum 270 2.2.2.5 They would clean up

[P_1]

Well there are really two main different ram markets TCCD and BH-5. These are quite different one relies on timings another on raw speed and for some processors TCCD will be better, and others BH-5. I really don't think that selling something in between will help because if you are trying to reach a high HTT with a low multi you will just grab yourself some nice tccd chips and overclock. That is good mostly for amd processors because you can not really run 300+ htt 24/7 on an intel platform cause that would force you to a 50% oc, which most chips can not handle, so bh-5 tends to be more popular with intel. As far as I know both the chip types are about equally used on amd platforms. If you made something in between it will not be bought much because it will no longer have a specific use and will not be better than either type. You should just come up with your own TCCD and BH-5 and try to beat the current chips by raw speed or timings, than you will definately get some sales.

no not really, what if his company developed chips that are capable of 2-2-2 timings at 300mhz while doing even tighter timings at lower hz? That would simply be amazing.

[trakslacker]

^ Amazing, but highly unlikely. Would take a ton of R&D.

If your company is actually going through with developing and producing their own RAM and PCBs, they need to shoot for something that can run high(270-300) at tight timings(2-2-2-x), at low volts. Unfortunately, this is seemingly an impossible situation in today's RAM market. But if it could happen and you produced a chip rated to ~270 @ 2-2-2 and was that was binned to go even higher all at low volts, you would make a killing. B/c then even joe schmoe could grab a pair and get great performance.(This is generally not the case at all with VX, BH-5, etc. due to volts needed.)

my .02

[P_1]

^ Amazing, but highly unlikely. Would take a ton of R&D.

If your company is actually going through with developing and producing their own RAM and PCBs, they need to shoot for something that can run high(270-300) at tight timings(2-2-2-x), at low volts. Unfortunately, this is seemingly an impossible situation in today's RAM market. But if it could happen and you produced a chip rated to ~270 @ 2-2-2 and was that was binned to go even higher all at low volts, you would make a killing. B/c then even joe schmoe could grab a pair and get great performance.(This is generally not the case at all with VX, BH-5, etc. due to volts needed.)

my .02

they should make it a combination of winbond/tccd by making it scale higher at tighter timings and even higher at high volts. Or make them rated at 1-1-1 @ 200mhz and by "loosening" them to 2-2-2 they can hit 270-300 mhz
and all done at a low low voltage of 2.6v and has lots of headroom to go higher with more voltage.

Also the chips should be as cool as possible so that we dont need extra fans in our cases just to cool the ram. Also dont forget that heatspreaders = bad, if you wanna make your ram look nice and cool find a new solution other than these crappy heatspreaders.(maybe use better material and epoxy them onto the mem modules on instead of using thermal tape, would be MUCH better)

[WeStSiDePLaYa]

wow, why doesnt everyone just ask for magic ram that will do 1.0-1-1-4 at 300mhz with 2.6vdimm? and it will run super cool, and cost 50$/GB, and it will do your taxs for you too.

[kristos]

^ Amazing, but highly unlikely. Would take a ton of R&D.

not really, the technology is there, production would just cost to much for it to be profitable.

[metro.cl]

we need at least 300 2-2-2-5 with safe voltages (no more than 3.3v) more would be awsome but 300 is ok

[jiff]

Its all well and good saying no more 3.3+volts, but who would say no to a set of sticks that did 270+ 2.2.2.5 with 3.3volts guranteed

[kromosto]

nice job

[P_1]

not really, the technology is there, production would just cost to much for it to be profitable.

I doubt that... it costs winbond very CHEAP to make their current utt stuff(like 10 dollars for a GB of chips or something) so even if the chips cost a lot more it would just mean companies selling sticks at 200+ would just be making less profit then usual.(I dunno about tccd chips but the companies are making tons off the utt stuff because of how cheap it costs) If the chips cost even 100 a GB (10fold of what utt costs right now) then the companies would still make money, just less of it, or even the same since everyone would be buying their stuff instead of the competitions stuff.

[GazC]

Hey guys...I'm not sure if this belongs in this forum but I think it's the closest to the topic at hand. I just got hired by a local ram company who wants to enter the gaming/overclocking market to overclock and beat the crap out of their ram. And I'm supposed to give them feedback.

Now I know it won't be BH-5 or TCCD, but just wondering what I should tell them in terms of a balance of timings vs. raw speed.

Any input from you guys would be appreciated since you guys are all extreme here. Thanks again!

Why not use the UTT? It's one of the cheapest ICs on the market and the BP PCB is cheaper than the JEDEC PCB. All you guys would need to do is refine the speed binning like the others have said.

Tbh, I'm a little sick of guys who get to test Ram that does a good deal better than it's mass production brethren. It's happening a little TOO often now a days. Time was that it used to be review samples in online product reviews. Now we get guys opening threads in forums to pimp their uber ram in. I would have a lot of time for a company whos retail product matched the performance of its review samples. As opposed to giving cherries out to selected people.

[jiff]

Tbh, I'm a little sick of guys who get to test Ram that does a good deal better than it's mass production brethren. It's happening a little TOO often now a days. Time was that it used to be review samples in online product reviews. Now we get guys opening threads in forums to pimp their uber ram in. I would have a lot of time for a company whos retail product matched the performance of its review samples. As opposed to giving cherries out to selected people.

Well Said

[Napoleonic]

DON'T FORGET ABOUT DDR2 (for M2 soket ha ) and beyond.....

[dippyskoodlez]

Why not use the UTT? It's one of the cheapest ICs on the market and the BP PCB is cheaper than the JEDEC PCB. All you guys would need to do is refine the speed binning like the others have said.

Tbh, I'm a little sick of guys who get to test Ram that does a good deal better than it's mass production brethren. It's happening a little TOO often now a days. Time was that it used to be review samples in online product reviews. Now we get guys opening threads in forums to pimp their uber ram in. I would have a lot of time for a company whos retail product matched the performance of its review samples. As opposed to giving cherries out to selected people.

Thats why the stuff I'm reviewing, I'm buying