Opinions on RAM please

[Rocky_53 1,579]

Has anyone got any advice or opinions on Patriot Viper Steel 2x 16GB DDR4-3200 RAM? I'm upgrading my setup slowly and as I also fly DCS, upgrading the RAM will be the next step, followed by a new CPU. Thanks fellas.

[micstatic 4,319]

I don't think you will see much of an improvement.  You already have 3200 ram.  Guess it depends on the CL you have now.  Though if you are super committed to upgrading the ram, now may be a good time.  Most suggest prices will be going up. 

[asgnp 27]

I tested 3200 c16 memory at 3400 c16, 3600 c16. There was barely any difference.

However i did see a 4-6% difference on going from 2x8gb to 4x8gb. This is due to dram interleaving.

And yes, prices have already begun to rise, so maybe a good time to upgrade memory and storage.

[Christopher Low 5,352]

I am no expert, but replacing 3200Mhz RAM with more 3200Mhz RAM sounds like wasted money to me, although I am quite prepared to be proved wrong if someone could explain it to me.

[Bob Scott 7,719]

Any time I see someone ask a question about RAM performance that only mentions the frequency, I sigh.  We've been through this a zillion times here already...memory performance is best measured by the true latency--the time it takes from a read request until the data is available on the bus.  It is a function of both memory clock frequency and CAS latency.  You can have memory operating at a very high frequency that has high CAS latency, and is therefore no better (or even worse) in performance that another memory set that runs at a lower frequency.  Referring to frequency without mention of CL usually means that this is not understood, and therefore the potential buyer is a ripe target for the sales ploys that would have you believe that 3600MHz RAM is just better than 3200MHz RAM, regardless of CL. 

It's like asking what the better deal is...a bottle of beer for $2 or a bottle of beer for $3--with no mention of the size of the bottle. 

See the discussion at

Regards

[Mikelab6 89]

Hi,

I think Howard is asking about the brand and model itself, not the speed. He is upgrading from 16 Gb to 32 GB, he is not upgrading for the speed.

Mike

[G-RFRY 1,710]

Latency is a factor in Ram performance the lower the better also more expensive. 

[SteveW 1,792]

So long as latency is not too long the memory will work as fast as the locations can be cycled to. In effect even less latency doesn't factor since it won't be cycled to any earlier anyway. Speed is more important.

[Rocky_53 1,579]

5 hours ago, Mikelab6 said:

Hi,

I think Howard is asking about the brand and model itself, not the speed. He is upgrading from 16 Gb to 32 GB, he is not upgrading for the speed.

Mike

Correct.Thank you Mike, for a minute I thought people were not reading properly!    opinions on Patriot Viper Steel 2x 16GB DDR4-3200 RAM  

Edited February 8, 2020 by Rockliffe

[Rocky_53 1,579]

6 hours ago, w6kd said:

Any time I see someone ask a question about RAM performance that only mentions the frequency, I sigh.  We've been through this a zillion times here already...memory performance is best measured by the true latency--the time it takes from a read request until the data is available on the bus.  It is a function of both memory clock frequency and CAS latency.  You can have memory operating at a very high frequency that has high CAS latency, and is therefore no better (or even worse) in performance that another memory set that runs at a lower frequency.  Referring to frequency without mention of CL usually means that this is not understood, and therefore the potential buyer is a ripe target for the sales ploys that would have you believe that 3600MHz RAM is just better than 3200MHz RAM, regardless of CL. 

It's like asking what the better deal is...a bottle of beer for $2 or a bottle of beer for $3--with no mention of the size of the bottle. 

See the discussion at

Regards

 

Taken from the DCS website relating to PC specification requirements:-        RAM: 16 GB (32 GB for heavy missions)

[SteveW 1,792]

https://www.crucial.com/usa/en/memory-performance-speed-latency

 

[Bob Scott 7,719]

40 minutes ago, SteveW said:

https://www.crucial.com/usa/en/memory-performance-speed-latency

 

Well...that Crucial article has been around for a while, and also it has been lampooned as a sales pitch for high-frequency memory disguised as a scientific fact sheet.  They throw out some facts, and then make a leap to draw conclusions not supported by the facts they laid down to begin with.  Consider the source...an outfit that profits from selling high-latency memory labelled as performance memory.  They sell some good stuff, and they also market some pretty mediocre memory as something it isn't.  Caveat emptor.

 

[SteveW 1,792]

10 minutes ago, w6kd said:

Well...that Crucial article has been around for a while, and also it has been lampooned as a sales pitch for high-frequency memory disguised as a scientific fact sheet.  They throw out some facts, and then make a leap to draw conclusions not supported by the facts they laid down to begin with.  Consider the source...an outfit that profits from selling high-latency memory labelled as performance memory.  They sell some good stuff, and they also market some pretty mediocre memory as something it isn't.  Caveat emptor.

 

So you say. What's the proof?

[SteveW 1,792]

The memory location has to be ready when read so if the latency was too long it wouldn't work. So what I think they are referring to is that frequency counts. If it handles the frequency, then it handles it with the latency measured at that frequency.

Edited February 9, 2020 by SteveW

[SteveW 1,792]

So it might be that paying for latency at the same frequency when it makes no odds is the wool pulled over the other's eyes.