Will expensive ram give me more performance than cheap ram?

[manikyath]

1 minute ago, stealth80 said:

 

was only GTAV and RotTR and only 4k using 2x 970's

Also http://www.tomshardware.co.uk/answers/id-2958259/rise-tomb-raider-high-vram.html

i'm sure you're saying the numbers you're seeing, i'm just questioning if the numbers you're seeing are normal

[stealth80]

1 minute ago, manikyath said:

i'm sure you're saying the numbers you're seeing, i'm just questioning if the numbers you're seeing are normal

No not at all, as I said it was one specific set up on 2 titles only, it had to down to vram issues. I also have zero pagefile cause of 32gb ram

[stealth80]

3 hours ago, Cantaloupeman said:

https://pcpartpicker.com/list/tv2P7h So this is my build that id like to put together. I will most likely put in an rx480 or something around that performance. The ram however sort of caught me up. The good 3000mhz ram 8gb from corsair is the cheapest 8gb 3000mhz ram available. But it's 60$. Could i just get generic 8gb ddr4 and call it good? Or does ram speed really make a difference?

Something like this:

 

PCPartPicker part list / Price breakdown by merchant

CPU: Intel Core i5-7500 3.4GHz Quad-Core Processor  ($201.89 @ B&H)
Motherboard: Gigabyte GA-H270-HD3 ATX LGA1151 Motherboard  ($101.98 @ Newegg)
Memory: Avexir Core Series 16GB (2 x 8GB) DDR4-2400 Memory  ($79.99 @ Newegg)
Power Supply: EVGA SuperNOVA G2 550W 80+ Gold Certified Fully-Modular ATX Power Supply  ($78.89 @ OutletPC)
Total: $462.75
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2017-01-16 13:48 EST-0500

 

Don't skrimp on the PSU, 2400mhz is the sweet spot for ram, no need for faster unless you can pay less

[JoostinOnline]

1 hour ago, stealth80 said:

No not at all, as I said it was one specific set up on 2 titles only, it had to down to vram issues. I also have zero pagefile cause of 32gb ram

To be clear, I wasn't questioning your honesty.  I just wanted to see wtf was going on in Task Manager.  Like if it had a secondary process or something.

[stealth80]

47 minutes ago, JoostinOnline said:

To be clear, I wasn't questioning your honesty.  I just wanted to see wtf was going on in Task Manager.  Like if it had a secondary process or something.

Nah was nothing like that, I just wanna see it framed

 

Yes it definitely the games, only those 2, only on the 970 SLI and only at 4k, I checked all that at the time. I've never had it since I upgraded to the 1080

[JoostinOnline]

3 minutes ago, stealth80 said:

Yes it definitely the games, only those 2, only on the 970 SLI and only at 4k, I checked all that at the time. I've never had it since I upgraded to the 1080

Wait, the increased RAM stopped when you upgraded your GPU?

[stealth80]

3 minutes ago, JoostinOnline said:

Wait, the increased RAM stopped when you upgraded your GPU?

yup, aint had it since the 970 SLI, hence why I said it must have been the lack of vram

[JoostinOnline]

53 minutes ago, stealth80 said:

yup, aint had it since the 970 SLI, hence why I said it must have been the lack of vram

Still seems super strange to me.  4GB is a lot of VRAM already.

46 minutes ago, wrathoftheturkey said:

An emphatic 

NO

If you're replying to the original post, you're wrong.  Better RAM means better performance.  Plenty of benchmarks will back that up.  Go through the thread, you'll find examples.

[stconquest]

1 hour ago, JoostinOnline said:

Still seems super strange to me.  4GB is a lot of VRAM already.

 

3.5GB.

 

1 hour ago, JoostinOnline said:

If you're replying to the original post, you're wrong.  Better RAM means better performance.  Plenty of benchmarks will back that up.  Go through the thread, you'll find examples.

Better RAM means higher overclocks and tighter timings, not better performance.  Two kits running at 3000MHz and 15CAS will act similarly.  The better kit will hit higher clocks and better timings when pushed.  Most users don't push their RAM.

[JoostinOnline]

1 minute ago, stconquest said:

3.5GB.

 

Better RAM means higher overclocks and tighter timings, not better performance.  Two kits running at 3000MHz and 15CAS will act similarly.  The better kit will hit higher clocks and better timings when pushed.  Most users don't push their RAM.

You can get higher frame rates.  On top of that, I've had computers run much slower when upgrading to cheaper RAM that had the same timings and frequency.  Not all RAM is created equally.

[HKZeroFive]

Wait, judging from this thread, do people seriously think that faster RAM makes little to no difference at all despite the overwhelming amount of evidence online? Yeesh.

 

The bottom line is this; the biggest benefit of faster memory is higher minimum framerates. However, if you're in a CPU-bound scenario/game where the CPU is essentially the bottleneck, you will see that faster memory boosts the average framerate.

 

In your case OP, you want a Z170 motherboard to take advantage of faster RAM. To be frank, your build is all over the place. An unlocked i5 6600K with a non-overclockable motherboard and you've cheaped out and put in a trash power supply (that scored a 5.5 on JG IIRC) which shouldn't be used for a build of this calibre.

 

1080p/Titan X Pascal OC	Core i7 7700K Stock	Core i7 7700K Stock	Core i7 7700K Stock	Core i7 7700K 4.8GHz	Core i7 7700K 4.8GHz	Core i7 7700K 4.8GHz
DDR4 Clock Speed	3000MHz	2400MHz	2133MHz	3000MHz	2400MHz	2133MHz
Assassin's Creed Unity, Ultra High, FXAA	132.2	129.3	125.3	132.9	129.8	126.6
Ashes of the Singularity, DX12, CPU Test	41.9	38.2	36.2	44.1	40.4	37.4
Crysis 3, Very High, SMAA T2x	138.2	135.0	132.8	145.5	140.7	138.0
The Division, Ultra, SMAA	133.8	133.6	132.1	133.9	133.2	132.6
Far Cry Primal, Ultra, SMAA	137.9	127.5	121.2	140.1	134.4	126.5
Rise of the Tomb Raider DX12, Very High, SMAA	126.5	114.9	108.1	131.0	121.1	111.0
The Witcher 3, Ultra, No Hairworks	139.4	125.7	116.6	145.2	130.8	120.2

I'm sorry, but the amount of ignorance here is staggering.

[stconquest]

2 hours ago, JoostinOnline said:

You can get higher frame rates.  On top of that, I've had computers run much slower when upgrading to cheaper RAM that had the same timings and frequency.  Not all RAM is created equally.

I find it hard to believe that IC (circuitry) and differing flash manufacturers can have that much of an impact at stock settings when those settings are similar.  I am not saying it is not possible, but do you have any other sources for data?

[JoostinOnline]

20 minutes ago, stconquest said:

I find it hard to believe that IC (circuitry) and differing flash manufacturers can have that much of an impact at stock settings when those settings are similar.  I am not saying it is not possible, but do you have any other sources for data?

On overclocking RAM, yes.  You'll find tons of videos for it.  Here's one:

Jump to 2:35 if you want to see the framerate differences.

 

On stock settings, no.  Believe me, I was shocked.  I switched out the sticks several times to test it.  I also put the RAM in another laptop and saw a speed boost too, so it wasn't just mine.  You can also use apps to benchmark the speed, like MaxxMem2.

[stconquest]

25 minutes ago, JoostinOnline said:

On overclocking RAM, yes.  You'll find tons of videos for it.  Here's one:

Jump to 2:35 if you want to see the framerate differences.

 

On stock settings, no.  Believe me, I was shocked.  I switched out the sticks several times to test it.  I also put the RAM in another laptop and saw a speed boost too, so it wasn't just mine.  You can also use apps to benchmark the speed, like MaxxMem2.

I know about frequency and timing benefits.  I am distinctly interested in seeing the difference IC and flash make when the frequency/timings are the same.  Sounds interesting.

[JoostinOnline]

43 minutes ago, stconquest said:

I know about frequency and timing benefits.  I am distinctly interested in seeing the difference IC and flash make when the frequency/timings are the same.  Sounds interesting.

Well I work on computers for a living, so I'll try to find some equal RAM with the same timings and frequency to compare. 

 

If it matters, the brand I used that ended up slower was Crucial. 

[stconquest]

16 minutes ago, JoostinOnline said:

Well I work on computers for a living, so I'll try to find some equal RAM with the same timings and frequency to compare. 

 

If it matters, the brand I used that ended up slower was Crucial. 

If you can definitively prove it, a lot of members here would love to see it; me too of course.

 

@MageTank  Sorry for the bother but this might be interesting.  Does IC, flash, etc make a difference to raw performance at similar clocks?  I assumed all the manufacturers would have equal performance provided the rated parts were working properly. 

 

Example:  Does 3000MHz RAM at 15 CAS perform the same at stock speeds across all manufacturers (excluding any longevity factor of the kit)?

[MageTank]

14 minutes ago, stconquest said:

If you can definitively prove it, a lot of members here would love to see it; me too of course.

 

@MageTank  Sorry for the bother but this might be interesting.  Does IC, flash, etc make a difference to raw performance at similar clocks?  I assumed all the manufacturers would have equal performance provided the rated parts were working properly. 

 

Example:  Does 3000MHz RAM at 15 CAS perform the same at stock speeds across all manufacturers (excluding any longevity factor of the kit)?

For the most part, yes. The major difference between the IC's (aside from obvious overclocking/voltage tolerances) is their individual preferences for certain tertiary timings to be tighter or looser than others. You might have Samsung IC's that get more bandwidth with specific tertiary timings being tighter, or Hynix IC's that get faster when the exact same tertiary timings are looser. 

 

You also have to understand that the ram itself isn't the only factor involved here. Your CPU's IMC plays the largest part in tertiary timings. Your motherboard (more specifically, memory power delivery and trace topology) plays a huge part in RTL/IO-L offsets, arguably the most important settings when determining speed/latency once all else is tuned. While some might see a big difference between IC's on a specific board/CPU combo, others might not see it at all, because their IMC and board trained the timings completely differently. 

 

I know for certain if you compare tertiary training on Asrock and ASUS boards, ASUS completely outclasses Asrock in that regard. You can have the exact same kit and CPU's in both boards, and the ASUS board might have more raw bandwidth and lower latency out of the box simply due to it's superior training. Once you manually tweak on both boards however, that advantage goes away. Overall IMC quality, IC quality, and superior trace topology plays the biggest roles in that scenario.

[stconquest]

21 minutes ago, MageTank said:

For the most part, yes. The major difference between the IC's (aside from obvious overclocking tolerances) is their individual preferences for certain tertiary timings to be tighter or looser than others. You might have Samsung IC's that get more bandwidth with specific tertiary timings being tighter, or Hynix IC's that get faster when the exact same tertiary timings are looser. 

 

You also have to understand that the ram itself isn't the only factor involved here. Your CPU's IMC plays the largest part in tertiary timings. Your motherboard (more specifically, memory power delivery and trace topology) plays a huge part in RTL/IO-L offsets, arguably the most important settings when determining speed/latency once all else is tuned. While some might see a big difference between IC's on a specific board/CPU combo, others might not see it at all, because their IMC and board trained the timings completely differently. 

 

I know for certain if you compare tertiary training on Asrock and ASUS boards, ASUS completely outclasses Asrock in that regard. You can have the exact same kit and CPU's in both boards, and the ASUS board might have more raw bandwidth and lower latency out of the box simply due to it's superior training. Once you manually tweak on both boards however, that advantage goes away. Overall IMC quality, IC quality, and superior trace topology plays the biggest roles in that scenario.

Sorry, had to google some stuff.  Your info is above my pay grade.

 

Assuming IOL, Round trip, IMC(CPU) is all the same from setup to setup, popping in different RAM at similar frequency/CAS could result in some performance variant as to the way manufacturer's flash chips act to changes in tertiary timings?

 

Example:  Samsung flash + IC will perform differently to Hynix +IC because of the way tertiary timings are handled by the setup (CPU-IMC/mobo)?  Some kits will be better off with certain setups because of the way the system deals out power, auto setting to timings (tertiary), etc...?

[MageTank]

1 minute ago, stconquest said:

Sorry, had to google some stuff.  Your info is above my pay grade.

 

Assuming I/O, Round trip, IMC(CPU) is all the same from setup to setup, popping in different RAM at similar frequency/CAS could result in some performance variant as to the way manufacturer's flash chips act to changes in tertiary timings?

 

Example:  Samsung flash + IC will perform differently to Hynix +IC because of the way tertiary timings are handled by the setup (CPU-IMC/mobo)?  Some kits will be better off with certain setups because of the way the system deals out power, auto setting to timings (tertiary), etc...?

For real world, the variance is so slight that you simply will not notice it. No person, myself included, can tell the difference between Samsung, Micron, or Hynix in gaming. As long as their raw bandwidth and latency is similar (and believe me, at the same clock/primaries without fine tuning the tertiary timings, they will be) you will get the exact same experience. 

 

The quality of the IC's only really matter if you are trying to push for exotic clock speeds beyond the means of anything practical. You have Samsung on the very high end of consumer memory (4266mhz) because it does extremely well on air. You have Hynix (MFR) which is used in the LN2 community simply because it tolerates high volts without any cold bug issues. You have Micron, because... it's cheap. You will notice most single rank, sub-3000mhz kits are almost always Micron. Kingston FuryX, anything with Crucials name attached to it, EVGA's cheapo DDR4, etc all use Micron, arguably the worst IC when manually overclocking, yet make no difference in gaming as long as you reach optimal bandwidth and latency in situations with CPU overhead.

 

That's basically the point. There is no one memory speed that will magically solve every problem you have. Certain configurations will benefit from faster memory while others may not. Having higher end GPU's for lower resolutions, you end up with more CPU overhead than you would with a more balanced setup. In that situation, getting the fastest possible ram makes sense. However, if your GPU is often the bottleneck, faster ram won't help averages much at all. Now, there are exceptions to this rule, depending on the title you are playing. Some games just have CPU overhead due to the way they are programmed. Perhaps certain parts of the game lack threaded optimizations, or are just too stressful on the CPU regardless of your setup. Faster ram will help, and this is why we say "Faster ram helps minimum framerates" because honestly, that's always the case, regardless of the GPU. 

[stconquest]

Thanks for the info once again Magetank. 

[MageTank]

7 minutes ago, stconquest said:

Thanks for the info once again Magetank. 

No problems. If it's any help, Hynix and Samsung both publish the whitesheets to their memory online. They even list specific timings that their IC's prefer. Was really helpful for me, as I was able to actually learn the function of tCKE and it's tolerances, lol. You will see however, that most of them are very similar, and even if you are pushing for the absolute best manual overclocks, the difference will likely be less than 5% difference at absolute best. 

[JoostinOnline]

2 hours ago, MageTank said:

For the most part, yes. The major difference between the IC's (aside from obvious overclocking/voltage tolerances) is their individual preferences for certain tertiary timings to be tighter or looser than others. You might have Samsung IC's that get more bandwidth with specific tertiary timings being tighter, or Hynix IC's that get faster when the exact same tertiary timings are looser. 

 

You also have to understand that the ram itself isn't the only factor involved here. Your CPU's IMC plays the largest part in tertiary timings. Your motherboard (more specifically, memory power delivery and trace topology) plays a huge part in RTL/IO-L offsets, arguably the most important settings when determining speed/latency once all else is tuned. While some might see a big difference between IC's on a specific board/CPU combo, others might not see it at all, because their IMC and board trained the timings completely differently. 

 

I know for certain if you compare tertiary training on Asrock and ASUS boards, ASUS completely outclasses Asrock in that regard. You can have the exact same kit and CPU's in both boards, and the ASUS board might have more raw bandwidth and lower latency out of the box simply due to it's superior training. Once you manually tweak on both boards however, that advantage goes away. Overall IMC quality, IC quality, and superior trace topology plays the biggest roles in that scenario.

The part about the motherboard mattering interests me. This was the first time I saw an actual drop in performance from upgrading RAM, and like I said, the latency and frequency were the same. It was on a Toshiba laptop. 

[stconquest]

1 hour ago, JoostinOnline said:

The part about the motherboard mattering interests me. This was the first time I saw an actual drop in performance from upgrading RAM, and like I said, the latency and frequency were the same. It was on a Toshiba laptop. 

Shot in the dark here but it might have had something to do with "training" whereas the system had trained the original RAM to function with tighter timings than advertised.  I am not even sure if this is plausible.  How big of a change was there?

 

http://www.tomshardware.com/reviews/ram-overclocking-guide,4693-3.html

Memory Training

There are multiple clocks in a system (CPU, IMC, memory, etc.), with a variable tick/tock initialization from start-up to start-up, a wide variety of signal pathways and variable environmental parameters, all of which combine to create a disparity (a skew) in the real arrival time of various signals at their destination. The pre-boot DDR calibration sequence introduces various delays between signals in order to achieve synchronicity. This is where DDR training kicks in; there are a number of patterns (either preset/provided by vendors, or custom-made) that test various signal/delay sets for the best possible ranges of these values. The accuracy of these delays determines the RTL/IOL, and ultimately influences memory performance. Since RTLs and IOLs are set at boot, training has a very real impact on the CAS latency.

 

Fast Boot settings either skip the memory training entirely, or use a very rough-and-ready form of training. While this is good enough for normal purposes, the best possible training sequence (determined from literature, or comparing the RTL/IOL values resulting from using each test, or in the absence of additional data, using the sequence that takes the longest time) should be used when fine-tuning memory parameters or benchmarking, because the variable signal/delay accuracy from a sub-par training regime makes parameter comparison questionable. Still, if enthusiasts are looking for moderate increases in memory performance, this step is generally optional.

[JoostinOnline]

1 minute ago, stconquest said:

Shot in the dark here but it might have had something to do with "training" whereas the system had trained the original RAM to function with tighter timings than advertised.  I am not even sure if this is plausible.  How big of a change was there?

This was probably 3 years ago, but it was big enough that I immediately noticed it.  I just didn't understand how there could be such a large difference, especially when increasing the capacity.  Like I said, I swapped the RAM back and forth several times to make sure I wasn't imagining it.

 

This Christmas I upgraded to this set (which was much cheaper at the time BTW), and I didn't notice any performance changes, other than SuperFetch working for apps that I haven't opened in a longer point in time.

[stconquest]

2 minutes ago, JoostinOnline said:

This was probably 3 years ago, but it was big enough that I immediately noticed it.  I just didn't understand how there could be such a large difference, especially when increasing the capacity.  Like I said, I swapped the RAM back and forth several times to make sure I wasn't imagining it.

 

This Christmas I upgraded to this set (which was much cheaper at the time BTW), and I didn't notice any performance changes, other than SuperFetch working for apps that I haven't opened in a longer point in time.

I am assuming the kit you replaced the original set with was a dual channel kit?  It crossed my mind earlier, but I figured you probably would have accounted for that.  If you went from two sticks to a single stick, there would be a loss of bandwidth.