GEIL memory, opinions?

[phillyTIM]

Hey everyone: I'm considering some RAM manufactured by a company called GEIL; what's your opinion/experience with this brand of memory?

Thanks much for sharing!

 

[Zap]

RAM is a commodity. Buy what suits your need at a price point you're willing to pay.

Geil has some interesting low voltage modules.

 

[taltamir]

they make ram, it works.
I use it in several of machines and I haven't had problems. (you could say that for pretty much any type of ram)

 

[YBS1]

The only time I've used Geil I ended up returning it as it would not run the speed/timings it was rated for. This was years ago though and a single sample, so take that for what it's worth. Personally I'm a big fan of Mushkin Redline modules.

 

[taltamir]

The only time I've used Geil I ended up returning it as it would not run the speed/timings it was rated for. This was years ago though and a single sample, so take that for what it's worth. Personally I'm a big fan of Mushkin Redline modules.

to be fair I saw that once or twice with most names in ram.. OCZ, corsair, and a few others.
1. There are defective parts from each manufacturer, it happens.
2. Some specific models are just claiming too much and can't deliver, happens to all of them.,
3. ALL RAM with very few exceptions (which are model specific, not manufacturer specific) is advertised at values of serious overclocking... If the RAM says "DDR2-800 with 5-5-5-15" timings? It is going to probably be DDR2-800 7-7-7-18 @ default voltage or sometimes elevated voltage... you CAN overclock it to 5-5-5-15 by increasing the voltage (to what they tell you to)... check the fine print and look for VOLTAGE numbers.

the ram business is extremely cutthroat and I have yet to see a single ram company that hasn't done the above three... except many some "value" rams which are as cheap as possible, as low performance as possible, and actually admit it.

You should note though, that:
1. Many motherboards ARE capable of pushing that extra voltage those rams require. But some don't, your mobo should allow at least a 0.3 volt overvolt on the ram.
2. Usually they ARE able to sustain the advertised overclock speed at the advertised voltage.
3. RAM speeds are the least significant factor of your total system speed. As such, running at slightly worse timing is not going to significantly alter your overall performance.

PS: a little bit about speed. DDR1/2/3 means the kind of ram. -800 is the MHZ speed of that ram, the higher it is, the faster it is. 5-5-5-18 is 4 common timings, the LOWER those are the faster it is. Note that they should be divided by the DDR number if you want to compare them to each other.
that is, DDR1 timing of 2.5-2.5-2.5-5 is an equal amount of milisecond delay as DDR2 5-5-5-10, which is equal to DDR3 7.5-7.5-7.5-15
For practical reasons, it cannot go beyond certain numbers...

Sometimes they add -1T or -2T at the end... that is another timing thing...

again, all of it is largely irrelevant... you buy the cheapest ram that works, put it in, and there you have it...

 

[fleshconsumed]

Had couple of DDR modules back in the day because they had decent specs and price. Worked fine without any problems.

 

[fffblackmage]

I have a set of Geil 2x1GB DDR2-800 ram. They work and the only issue is incompatibility with a spare mobo I have.
-------

3. ALL RAM with very few exceptions (which are model specific, not manufacturer specific) is advertised at values of serious overclocking... If the RAM says "DDR2-800 with 5-5-5-15" timings? It is going to probably be DDR2-800 7-7-7-18 @ default voltage or sometimes elevated voltage... you CAN overclock it to 5-5-5-15 by increasing the voltage (to what they tell you to)... check the fine print and look for VOLTAGE numbers.

I'm not sure if changing the ram timings is the same as overclocking. You haven't change the clockspeed if you're just changing timings....

RAM speeds are the least significant factor of your total system speed. As such, running at slightly worse timing is not going to significantly alter your overall performance.

Timings and clockspeed are not the same thing. However, neither are usually significant in total system performance, timings being less significant than clock frequency probably.

PS: a little bit about speed. DDR1/2/3 means the kind of ram. -800 is the MHZ speed of that ram, the higher it is, the faster it is. 5-5-5-18 is 4 common timings, the LOWER those are the faster it is. Note that they should be divided by the DDR number if you want to compare them to each other.
that is, DDR1 timing of 2.5-2.5-2.5-5 is an equal amount of milisecond delay as DDR2 5-5-5-10, which is equal to DDR3 7.5-7.5-7.5-15
For practical reasons, it cannot go beyond certain numbers...

The timing numbers represent the number of clock cycle delays. You can only calculate the actual delay in time only if you have a clock frequency to work with. Therefore, timings like 2.5-2.5-2.5-5 for DDR may or may not be equal to 5-5-5-15 for DDR2. Those timings are equal only for DDR-400 and DDR2-800.

Also, ram cycles are measured in nanoseconds, not milliseconds. That's because a clock cycle of 800MHz would mean a single cycle lasts only 1.25 nanoseconds (1/800MHz). Multiply that with the timing of 5 and you get a 6.25 ns delay for DDR2-800.

There's no multiply or dividing with the number that comes after DDR. That only indicates the generation of DDR technology.

DDR-400 with CAS2.5
(1/400000000 Hz) * 2.5 = 6.25 ns

DDR2-800 with CAS5
(1/800000000 Hz) * 5 = 6.25 ns

DDR3-1600 with CAS10
(1/1600MHz) * 10 = 6.25 ns

However, change the frequency and keep the same timings and you'd get different numbers for delay.

DDR2-1000 with CAS5
(1/800000000 Hz) * 5 = 5 ns

 

[taltamir]

I'm not sure if changing the ram timings is the same as overclocking. You haven't change the clockspeed if you're just changing timings....

What would you call it? under-timing? by convention it is called overclocking

Timings and clockspeed are not the same thing. However, neither are usually significant in total system performance, timings being less significant than clock frequency probably.

Correct, but they have about the same (minute) effect

The timing numbers represent the number of clock cycle delays. You can only calculate the actual delay in time only if you have a clock frequency to work with. Therefore, timings like 2.5-2.5-2.5-5 for DDR may or may not be equal to 5-5-5-15 for DDR2. Those timings are equal only for DDR-400 and DDR2-800.

Which is the norm... DDR-400, DDR2-800, DDR3-1600 are the "standard" frequencies.

Also, ram cycles are measured in nanoseconds, not milliseconds. That's because a clock cycle of 800MHz would mean a single cycle lasts only 1.25 nanoseconds (1/800MHz). Multiply that with the timing of 5 and you get a 6.25 ns delay for DDR2-800.

You are correct, although technically its still an amount of time in miliseconds. 1 nanosecond = 0.000001 miliseconds

There's no multiply or dividing with the number that comes after DDR. That only indicates the generation of DDR technology.

Now you are just being condescending; and what do you think I am saying here exactly.
I clearly indicated that it is a method of comparing the relative "speed" of timings. and you get roughly the same result with an easier to remember/explain formula. Perhaps I should have, for completeness, included the exact forumla. but "divide out the timing numbers by generation of DDR and compare" works out well enough.

 

[fffblackmage]

I clearly indicated that it is a method of comparing the relative "speed" of timings. and you get roughly the same result with an easier to remember/explain formula. Perhaps I should have, for completeness, included the exact forumla. but "divide out the timing numbers by generation of DDR and compare" works out well enough.

DDR-400 with CAS 2.5 has a time delay of 6.25 ns.
DDR2-800 with CAS 5 has a time delay of 6.25 ns.
DDR3-1600 with CAS 10 has a time delay of 6.25 ns.

DDR3-1600 with CAS 7.5 would have a time delay of 4.69 ns.

Obviously, DDR3-1600 w/CAS 7.5 isn't consistent with being equal to the delay times of DDR-400 w/CAS 2.5 and DDR2-800 w/CAS 5. Thus, I'm saying divide by the number of DDR generation doesn't work at all.

 

[phillyTIM]

So what you guys are saying is that GEIL brings up no red flags at all, based upon your experience?

 

[taltamir]

So what you guys are saying is that GEIL brings up no red flags at all, based upon your experience?

yes