Jump to content

RAM Speeds vs Timings/Latency

Char170

I'm looking to build a new higher end pc and I want to understand RAM specs better.

 

What is the differences that I'd see between two types of RAM?

 

I have been looking at Corsair Vengeance 2x16 @ DDR4-4000 with timings of 18-22-22-42

 

But that has higher timings compared to something like:

G.Skill Trident Z 2x16 @ DDR4-3866 with timings of 18-18-18-38  (and about $35 cheaper)

 

So what is the differences here. What programs would benefit from higher frequency and which from lower timings?

 

(For reference this is the build I am currently looking for)

Link to comment
Share on other sites

Link to post
Share on other sites

Frequency affects bandwidth significantly and timings somewhat little, but combination of frequency and timings give you latency. Games care about latency a lot on Ryzen, but only some work (like crypto mining for example) care about bandwidth that much (above a certain point of course, say 3200MHz).

 

These two kit could be different in model of memory in it, but 3866 18-18-18 isnt tight enough that I can guarantee it will be anything certain.

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

Timings and freq are linked. They are not independent. You can do a simple equation to determine the real latency based on the freq and timings.

 

For the top kit:

 

18CL/(4000MHz/2) = 9.0ns CAS latency

 

For the bottom kit:

 

18CL/(3866MHz/2) = 9.3ns CAS latency

 

Aside from that, you have to think about your preferences and expectations. The best option for someone who does not want to have to fiddle with memory timings and frequencies and voltages would be to get a simple good kit of 3600Mhz CL16 RAM like these:

 

$165:

https://pcpartpicker.com/product/w3FKHx/gskill-trident-z-neo-32-gb-2-x-16-gb-ddr4-3600-memory-f4-3600c16d-32gtznc

 

If you think you might want to try to OC the memory on your own at a later date then maybe something like this would be better:

 

$235

https://pcpartpicker.com/product/6z6qqs/gskill-trident-z-rgb-32-gb-2-x-16-gb-ddr4-3600-memory-f4-3600c16d-32gtzr

 

If you want the fastest RAM and you want to customize your voltages and timings then you might even try some more expensive RAM like this:

 

$380

https://pcpartpicker.com/product/8jZFf7/gskill-tridentz-rgb-32gb-4-x-8gb-ddr4-4000-memory-f4-4000c17q-32gtzr

 

Honestly, the performance doesn't change much past 3200Mhz, so with your build and level of knowledge I'd recommend something more like the first kit I linked.

 

 

BabyBlu (Primary): 

  • CPU: Intel Core i9 9900K @ up to 5.3GHz, 5.0GHz all-core, delidded
  • Motherboard: Asus Maximus XI Hero
  • RAM: G.Skill Trident Z RGB 4x8GB DDR4-3200 @ 4000MHz 16-18-18-34
  • GPU: MSI RTX 2080 Sea Hawk EK X, 2070MHz core, 8000MHz mem
  • Case: Phanteks Evolv X
  • Storage: XPG SX8200 Pro 2TB, 3x ADATASU800 1TB (RAID 0), Samsung 970 EVO Plus 500GB
  • PSU: Corsair HX1000i
  • Display: MSI MPG341CQR 34" 3440x1440 144Hz Freesync, Dell S2417DG 24" 2560x1440 165Hz Gsync
  • Cooling: Custom water loop (CPU & GPU), Radiators: 1x140mm(Back), 1x280mm(Top), 1x420mm(Front)
  • Keyboard: Corsair Strafe RGB (Cherry MX Brown)
  • Mouse: MasterMouse MM710
  • Headset: Corsair Void Pro RGB
  • OS: Windows 10 Pro

Roxanne (Wife Build):

  • CPU: Intel Core i7 4790K @ up to 5.0GHz, 4.8Ghz all-core, relidded w/ LM
  • Motherboard: Asus Z97A
  • RAM: G.Skill Sniper 4x8GB DDR3-2400 @ 10-12-12-24
  • GPU: EVGA GTX 1080 FTW2 w/ LM
  • Case: Corsair Vengeance C70, w/ Custom Side-Panel Window
  • Storage: Samsung 850 EVO 250GB, Samsung 860 EVO 1TB, Silicon Power A80 2TB NVME
  • PSU: Corsair AX760
  • Display: Samsung C27JG56 27" 2560x1440 144Hz Freesync
  • Cooling: Corsair H115i RGB
  • Keyboard: GMMK TKL(Kailh Box White)
  • Mouse: Glorious Model O-
  • Headset: SteelSeries Arctis 7
  • OS: Windows 10 Pro

BigBox (HTPC):

  • CPU: Ryzen 5800X3D
  • Motherboard: Gigabyte B550i Aorus Pro AX
  • RAM: Corsair Vengeance LPX 2x8GB DDR4-3600 @ 3600MHz 14-14-14-28
  • GPU: MSI RTX 3080 Ventus 3X Plus OC, de-shrouded, LM TIM, replaced mem therm pads
  • Case: Fractal Design Node 202
  • Storage: SP A80 1TB, WD Black SN770 2TB
  • PSU: Corsair SF600 Gold w/ NF-A9x14
  • Display: Samsung QN90A 65" (QLED, 4K, 120Hz, HDR, VRR)
  • Cooling: Thermalright AXP-100 Copper w/ NF-A12x15
  • Keyboard/Mouse: Rii i4
  • Controllers: 4X Xbox One & 2X N64 (with USB)
  • Sound: Denon AVR S760H with 5.1.2 Atmos setup.
  • OS: Windows 10 Pro

Harmonic (NAS/Game/Plex/Other Server):

  • CPU: Intel Core i7 6700
  • Motherboard: ASRock FATAL1TY H270M
  • RAM: 64GB DDR4-2133
  • GPU: Intel HD Graphics 530
  • Case: Fractal Design Define 7
  • HDD: 3X Seagate Exos X16 14TB in RAID 5
  • SSD: Inland Premium 512GB NVME, Sabrent 1TB NVME
  • Optical: BDXL WH14NS40 flashed to WH16NS60
  • PSU: Corsair CX450
  • Display: None
  • Cooling: Noctua NH-U14S
  • Keyboard/Mouse: None
  • OS: Windows 10 Pro

NAS:

  • Synology DS216J
  • 2x8TB WD Red NAS HDDs in RAID 1. 8TB usable space
Link to comment
Share on other sites

Link to post
Share on other sites

I have been wondering this myself. I have a 3000Mhz 15-17-17-35 32gb kit in my system, and it goes up to 3.2Ghz without timing changes just by me upping the multiplier a bit more. This is with the ram being run at 1.36v according to my motherboard's live bios sensor.

I don't know how much further I can realistically push the kit on voltage since I read somewhere that DDR4 doesn't like going above the 1.35 it must already be configured for. Although I can see the timings are relatively tight by default, I am worried an upgrade to Zen3 from my rather ram-apathetic 6700k would be squandering some performance if I can't push it a little further.

Link to comment
Share on other sites

Link to post
Share on other sites

@pipnina

 

I have my DDR4 running @ 1.45V daily. Up to 1.5V is fine for daily if there is a benefit. With good airflow and higher end dimms you could go even higher, maybe 1.55V if it provides better performance.

 

15-17-17-35 @ 3200MHz isn't really that tight. Pretty average imho. If you want to find out what your kit is capable of, you could start by setting the voltage to 1.45V, and loosening the timings to something like 18-20-20-40 (play around with it) and seeing how high you can get with freq. When you hit a wall you can try loosening the timings again. Once you find your max clock, you can work on tightening the timings. From there you would reduce the clocks and timings until you find the combination that gives you the best real latency as I calculated above. And then you could fiddle with reducing the voltage if you wanted to. 16-19-19-36 @ 3600MHz might be within reach with a voltage bump.

 

It's a lot of work and a lot of trial and error for not that much performance increase, though. At 3200MHz you'll be getting most of the performance out of Zen 3.

BabyBlu (Primary): 

  • CPU: Intel Core i9 9900K @ up to 5.3GHz, 5.0GHz all-core, delidded
  • Motherboard: Asus Maximus XI Hero
  • RAM: G.Skill Trident Z RGB 4x8GB DDR4-3200 @ 4000MHz 16-18-18-34
  • GPU: MSI RTX 2080 Sea Hawk EK X, 2070MHz core, 8000MHz mem
  • Case: Phanteks Evolv X
  • Storage: XPG SX8200 Pro 2TB, 3x ADATASU800 1TB (RAID 0), Samsung 970 EVO Plus 500GB
  • PSU: Corsair HX1000i
  • Display: MSI MPG341CQR 34" 3440x1440 144Hz Freesync, Dell S2417DG 24" 2560x1440 165Hz Gsync
  • Cooling: Custom water loop (CPU & GPU), Radiators: 1x140mm(Back), 1x280mm(Top), 1x420mm(Front)
  • Keyboard: Corsair Strafe RGB (Cherry MX Brown)
  • Mouse: MasterMouse MM710
  • Headset: Corsair Void Pro RGB
  • OS: Windows 10 Pro

Roxanne (Wife Build):

  • CPU: Intel Core i7 4790K @ up to 5.0GHz, 4.8Ghz all-core, relidded w/ LM
  • Motherboard: Asus Z97A
  • RAM: G.Skill Sniper 4x8GB DDR3-2400 @ 10-12-12-24
  • GPU: EVGA GTX 1080 FTW2 w/ LM
  • Case: Corsair Vengeance C70, w/ Custom Side-Panel Window
  • Storage: Samsung 850 EVO 250GB, Samsung 860 EVO 1TB, Silicon Power A80 2TB NVME
  • PSU: Corsair AX760
  • Display: Samsung C27JG56 27" 2560x1440 144Hz Freesync
  • Cooling: Corsair H115i RGB
  • Keyboard: GMMK TKL(Kailh Box White)
  • Mouse: Glorious Model O-
  • Headset: SteelSeries Arctis 7
  • OS: Windows 10 Pro

BigBox (HTPC):

  • CPU: Ryzen 5800X3D
  • Motherboard: Gigabyte B550i Aorus Pro AX
  • RAM: Corsair Vengeance LPX 2x8GB DDR4-3600 @ 3600MHz 14-14-14-28
  • GPU: MSI RTX 3080 Ventus 3X Plus OC, de-shrouded, LM TIM, replaced mem therm pads
  • Case: Fractal Design Node 202
  • Storage: SP A80 1TB, WD Black SN770 2TB
  • PSU: Corsair SF600 Gold w/ NF-A9x14
  • Display: Samsung QN90A 65" (QLED, 4K, 120Hz, HDR, VRR)
  • Cooling: Thermalright AXP-100 Copper w/ NF-A12x15
  • Keyboard/Mouse: Rii i4
  • Controllers: 4X Xbox One & 2X N64 (with USB)
  • Sound: Denon AVR S760H with 5.1.2 Atmos setup.
  • OS: Windows 10 Pro

Harmonic (NAS/Game/Plex/Other Server):

  • CPU: Intel Core i7 6700
  • Motherboard: ASRock FATAL1TY H270M
  • RAM: 64GB DDR4-2133
  • GPU: Intel HD Graphics 530
  • Case: Fractal Design Define 7
  • HDD: 3X Seagate Exos X16 14TB in RAID 5
  • SSD: Inland Premium 512GB NVME, Sabrent 1TB NVME
  • Optical: BDXL WH14NS40 flashed to WH16NS60
  • PSU: Corsair CX450
  • Display: None
  • Cooling: Noctua NH-U14S
  • Keyboard/Mouse: None
  • OS: Windows 10 Pro

NAS:

  • Synology DS216J
  • 2x8TB WD Red NAS HDDs in RAID 1. 8TB usable space
Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, HairlessMonkeyBoy said:

@pipnina

 

I have my DDR4 running @ 1.45V daily. Up to 1.5V is fine for daily if there is a benefit. With good airflow and higher end dimms you could go even higher, maybe 1.55V if it provides better performance.

 

15-17-17-35 @ 3200MHz isn't really that tight. Pretty average imho. If you want to find out what your kit is capable of, you could start by setting the voltage to 1.45V, and loosening the timings to something like 18-20-20-40 (play around with it) and seeing how high you can get with freq. When you hit a wall you can try loosening the timings again. Once you find your max clock, you can work on tightening the timings. From there you would reduce the clocks and timings until you find the combination that gives you the best real latency as I calculated above. And then you could fiddle with reducing the voltage if you wanted to. 16-19-19-36 @ 3600MHz might be within reach with a voltage bump.

 

It's a lot of work and a lot of trial and error for not that much performance increase, though. At 3200MHz you'll be getting most of the performance out of Zen 3.

Thanks for the tips! It might only ammount to 1-5% over my normal 3GHz XMP in games, but free performance is free performance!

I think the cooler I am thinking of for the Zen3 processor I'm planning on will probably help with airflow over the ram too (noctua u14s) since my dark rock 3 sucks air from over the top of the ram currently.

Link to comment
Share on other sites

Link to post
Share on other sites

5 minutes ago, pipnina said:

Thanks for the tips! It might only ammount to 1-5% over my normal 3GHz XMP in games, but free performance is free performance!

I think the cooler I am thinking of for the Zen3 processor I'm planning on will probably help with airflow over the ram too (noctua u14s) since my dark rock 3 sucks air from over the top of the ram currently.

You could get a lot more performance by tuning your secondary timings, most importantly tRFC(as low as you can) and TREFI(as high as you can up to 65535). Those two settings alone could make a big difference since they together control for how long and how often, respectively, your DRAM is refreshing rather than reading and writing. If you are interested, here is a great thread on memory OCing:

 

 

BabyBlu (Primary): 

  • CPU: Intel Core i9 9900K @ up to 5.3GHz, 5.0GHz all-core, delidded
  • Motherboard: Asus Maximus XI Hero
  • RAM: G.Skill Trident Z RGB 4x8GB DDR4-3200 @ 4000MHz 16-18-18-34
  • GPU: MSI RTX 2080 Sea Hawk EK X, 2070MHz core, 8000MHz mem
  • Case: Phanteks Evolv X
  • Storage: XPG SX8200 Pro 2TB, 3x ADATASU800 1TB (RAID 0), Samsung 970 EVO Plus 500GB
  • PSU: Corsair HX1000i
  • Display: MSI MPG341CQR 34" 3440x1440 144Hz Freesync, Dell S2417DG 24" 2560x1440 165Hz Gsync
  • Cooling: Custom water loop (CPU & GPU), Radiators: 1x140mm(Back), 1x280mm(Top), 1x420mm(Front)
  • Keyboard: Corsair Strafe RGB (Cherry MX Brown)
  • Mouse: MasterMouse MM710
  • Headset: Corsair Void Pro RGB
  • OS: Windows 10 Pro

Roxanne (Wife Build):

  • CPU: Intel Core i7 4790K @ up to 5.0GHz, 4.8Ghz all-core, relidded w/ LM
  • Motherboard: Asus Z97A
  • RAM: G.Skill Sniper 4x8GB DDR3-2400 @ 10-12-12-24
  • GPU: EVGA GTX 1080 FTW2 w/ LM
  • Case: Corsair Vengeance C70, w/ Custom Side-Panel Window
  • Storage: Samsung 850 EVO 250GB, Samsung 860 EVO 1TB, Silicon Power A80 2TB NVME
  • PSU: Corsair AX760
  • Display: Samsung C27JG56 27" 2560x1440 144Hz Freesync
  • Cooling: Corsair H115i RGB
  • Keyboard: GMMK TKL(Kailh Box White)
  • Mouse: Glorious Model O-
  • Headset: SteelSeries Arctis 7
  • OS: Windows 10 Pro

BigBox (HTPC):

  • CPU: Ryzen 5800X3D
  • Motherboard: Gigabyte B550i Aorus Pro AX
  • RAM: Corsair Vengeance LPX 2x8GB DDR4-3600 @ 3600MHz 14-14-14-28
  • GPU: MSI RTX 3080 Ventus 3X Plus OC, de-shrouded, LM TIM, replaced mem therm pads
  • Case: Fractal Design Node 202
  • Storage: SP A80 1TB, WD Black SN770 2TB
  • PSU: Corsair SF600 Gold w/ NF-A9x14
  • Display: Samsung QN90A 65" (QLED, 4K, 120Hz, HDR, VRR)
  • Cooling: Thermalright AXP-100 Copper w/ NF-A12x15
  • Keyboard/Mouse: Rii i4
  • Controllers: 4X Xbox One & 2X N64 (with USB)
  • Sound: Denon AVR S760H with 5.1.2 Atmos setup.
  • OS: Windows 10 Pro

Harmonic (NAS/Game/Plex/Other Server):

  • CPU: Intel Core i7 6700
  • Motherboard: ASRock FATAL1TY H270M
  • RAM: 64GB DDR4-2133
  • GPU: Intel HD Graphics 530
  • Case: Fractal Design Define 7
  • HDD: 3X Seagate Exos X16 14TB in RAID 5
  • SSD: Inland Premium 512GB NVME, Sabrent 1TB NVME
  • Optical: BDXL WH14NS40 flashed to WH16NS60
  • PSU: Corsair CX450
  • Display: None
  • Cooling: Noctua NH-U14S
  • Keyboard/Mouse: None
  • OS: Windows 10 Pro

NAS:

  • Synology DS216J
  • 2x8TB WD Red NAS HDDs in RAID 1. 8TB usable space
Link to comment
Share on other sites

Link to post
Share on other sites

I bought a hyperx 2 x 16GB 3600 CL17-21-21-21-39 (HX436C17FB3K2/32) sticks for Ryzen 5800x. I put them on ASUS Strix x570-E Gaming.

It seems there is a problem with correct XMP profile read. The best what I could configure was 3600Mhz @ CL18-21-21-41-87. Either this RAM is faulty or x570 has issues reading sticks correctly. Nominal timings were possible only at 2400Mhz. Any suggestions?

 

I have also tried same sticks on MSI x570 motherboard. Similar result. and different sticks Patriot Viper 2 x 8GB

 3733 @ CL17 which also gave CL18 after windows boot in CPU-Z.

 

If I buy different G.SKILL sticks which are 2x 16GB Will I also get CL18 instead of CL16?

 

image.png.5359570ea364796e1d99e2931591fc30.png

 

 

 

I am attaching also another screen with result after first run with all defaults except XMP profile

photo_2020-11-10_23-58-18.jpg.7a3e834df15ac238f3f584ee9743bc64.jpg

photo_2020-11-10_23-49-14.jpg

Link to comment
Share on other sites

Link to post
Share on other sites

Go to the sub timings menu and disable "Gear Down mode" and it will run the odd ball cas latency of 17. 

 

Link to comment
Share on other sites

Link to post
Share on other sites

23 minutes ago, ShrimpBrime said:

Go to the sub timings menu and disable "Gear Down mode" and it will run the odd ball cas latency of 17. 

 

Thanks a million that helped!

Link to comment
Share on other sites

Link to post
Share on other sites

Is Command Rate 2T expected for such timings?

NVM, configured it correctly now.Any tips to tune it more greatly appreciated. 

 

image.png.e2990103f172d8a281fb29b5c1d3dcd4.png

Link to comment
Share on other sites

Link to post
Share on other sites

33 minutes ago, MADRAFi said:

Is Command Rate 2T expected for such timings?

NVM, configured it correctly now.Any tips to tune it more greatly appreciated. 

 

image.png.e2990103f172d8a281fb29b5c1d3dcd4.png

Sure. The main timings you see here, tighten them. Do one at a time and test. Keep the 1T command rate. Be sure the memory voltage is set to 1.36v instead of 1.35v to accommodate for any v-droop. You also can disable power down mode, this will help prevent the memory from throttling/power saving.

Link to comment
Share on other sites

Link to post
Share on other sites

I will have to work on it. So far with 1T and these above settings I got blue screen immediately after heavy load from the game.

 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, MADRAFi said:

I will have to work on it. So far with 1T and these above settings I got blue screen immediately after heavy load from the game.

 

OK, switch back to 2T (which helps with stability in most cases) and test again. 

Link to comment
Share on other sites

Link to post
Share on other sites

8 hours ago, ShrimpBrime said:

OK, switch back to 2T (which helps with stability in most cases) and test again. 

I am back on 1T. your suggestion to make sure the is no variation with voltage and is set to 1.36 helped. Is it my impression or lower CL is better then higher memory speed?

Which timing should be changed when I increase memory speed from 15-17-17-36-69-1 on 3000 Mhz

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, MADRAFi said:

I am back on 1T. your suggestion to make sure the is no variation with voltage and is set to 1.36 helped. Is it my impression or lower CL is better then higher memory speed?

Which timing should be changed when I increase memory speed from 15-17-17-36-69-1 on 3000 Mhz

You would loosen the Cas latency, but try and keep the same timings and increase frequency until unstable. 

Link to comment
Share on other sites

Link to post
Share on other sites

10 minutes ago, ShrimpBrime said:

You would loosen the Cas latency, but try and keep the same timings and increase frequency until unstable. 

I am now back with 2T and XMP. Default XMP 3600MHz is unstable on 1T under load. Will play more with different XMP profiles (have 2)

Link to comment
Share on other sites

Link to post
Share on other sites

5 minutes ago, MADRAFi said:

I am now back with 2T and XMP. Default XMP 3600MHz is unstable on 1T under load. Will play more with different XMP profiles (have 2)

It will take some trial and error. But you're on the right track. Most anything that happens next is up to you. Small changes are best.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×