New RAM not running at MOBO supported speed

[mdklassen]

---

I can confirm the original RAM running at 2400 speed (via Speecy).

Original RAM Slot #2
    Type    Unknown
    Size    4096 MBytes
    Manufacturer    Hyundai Electronics
    Max Bandwidth    DDR4-2400 (1200 MHz)
    Part Number    HMA851U6AFR6N-UH

I added an 8GB stick of 2400 GSkill RAM and am only getting getting the following 2132 speed, which makes no sense!  Clearly the board and Orig stick run at 2400, so why does other 2400 RAM not run at that speed??  It even states its running in XMP!?!

Slot #1
    Type    Unknown
    Size    8192 MBytes
    Manufacturer    G.Skill
    Max Bandwidth    DDR4-2132 (1066 MHz)
    Part Number    F4-2400C15-8GIS
    SPD Ext.    XMP

---

I think that you may be misinterpreting the output from SPECCY.

For example, on my computer:

RAM

~~Memory slots
~~~Total memory slots~4
~~~Used memory slots~2
~~~Free memory slots~2

~~Memory
~~~Type~DDR3
~~~Size~8192 MBytes
~~~Channels #~Dual
~~~CAS# Latency (CL)~9 clocks
~~~RAS# to CAS# Delay (tRCD)~9 clocks
~~~RAS# Precharge (tRP)~9 clocks
~~~Cycle Time (tRAS)~24 clocks

So, take that "9-9-9-24", which is the current speed of the access to the RAM,
and compare with the potential "Timing table" for each stick:

~~Frequency~CAS# Latency~RAS# To CAS#~RAS# Precharge~tRAS~tRC~Voltage
~JEDEC #1~444.4 MHz~6.0~6~6~16~23~1.500 V
~JEDEC #2~592.6 MHz~8.0~8~8~22~30~1.500 V
~JEDEC #3~666.7 MHz~9.0~9~9~24~34~1.500 V

to determine the current usage of the RAM.  In my case, my stick is running at 667 MHz.

In your case, the G-SKILL RAM is reporting that it maxes-out at '2133', not '2400'.

See: https://www.gskill.com/en/product/f4-2400c15s-8gnt

Tested Speed: 2400 MHz 
Tested Latency: 15-15-15-35 
Tested Voltage: 1.20v 
Registered/Unbuffered: Unbuffered 
Error Checking: Non-ECC 
SPD Speed: 2133MHz

I don't know why the "tested speed" is higher than what the SPD (Serial Presence Detect) chip is reporting.

Is it capable of overclocking to 2400 MHz ???

[Big_Dave]

@mdklassen

I agree with your message #7.

[SemiTechie]

Thanks for the reply.  Admittedly I only have a cursor knowedge of how the RAM timings/latency work.  Sadly I can't OC the RAM as this is in an HP that appears to have the voltage locked down.  My item is actually the Aegis line, but other than the Model # differing slightly, looks like the specs are  exactly the same (Makes you wonder if GSkill just charges more because they gave it a cool name, slapped a colorful sticker on it, and called it Gaming RAM?!?).

https://www.gskill.com/en/product/f4-2400c15s-8gis

So as I interpret this moving forward, the RAM 'should/could' be capable of reaching the higher speed, but would need some tweeks/OC??  Seems a bit deceptive to me on GSkill's part.  Seems to run fine and benchmarks seem decent.  Is there any risk of problems down the line with the 2 speeds not matching up??  Or is my system automatically lowering the other Original 4GB stick down to 1066 to match the timing of this GSkill stick, and thus lowering the speed of the first?

GSkill has commented on NewEgg  board that the XMP profile needs to be set to achieve the 1200/2400 speed.  Clearly I think that should be prominantly mentioned in the Specs and not buried in a comment response.

Thanks!

[SemiTechie]

And in real terms... does any of this really matter?  Does the 2133 speed actually manifest itself in any way even perceptable to the end user??  I.E. is it perceptible at all to the user in most typical PC uses like Moderate gaming, Internet videos, Word/Excel/Etc. (not large-scale data crunching) ??

[Big_Dave]

@SemiTechie

The speed difference say between 2400 and 3200 would be noticeable with applications such as WinR and memory intensive applications.  Normal every day use wouldn't give you much of a noticeable "seat of the pants" feel.

Pick up a copy of MAXIMUM PC magazine, Spring 2018 edition as there is an in depth article on memory speeds.

[mdklassen]

1. Admittedly I only have a cursory knowedge of how the RAM timings/latency work. 

OK.

2. I can't OC the RAM, as this is in an HP that appears to have the voltage locked down.  

Overclocking may require both boosting the voltage and changing the "clock-timing", to make it "tick" faster.

I doubt that the BIOS on the HP motherboard supports either feature.

3. Makes you wonder if GSkill just charges more because they gave it a cool name, slapped a colorful sticker on it, and called it Gaming RAM?!?).

That RAM usually is covered by a heat-spreader, to conduct heat away from the RAM.

Again, bumping-up the voltage, to overclock, may generate more heat than "standard".

4. So as I interpret this, the RAM 'should/could' be capable of reaching the higher speed, but would need some tweaks/OC?? 

Agree.

5. Seems a bit deceptive to me on GSkill's part. 

See your comment in #1, above. 

6. Seems to run fine and benchmarks seem decent. 

Good.

7. Is there any risk of problems down the line with the 2 speeds not matching up?? 

No.  The motherboard is capable of handling mis-matched speeds of RAM.

8. Or is my system automatically lowering the other Original 4GB stick down to 1066,

to match the timing of this GSkill stick, and thus lowering the speed of the first?

Maybe.  It is possible that the motherboard runs the faster RAM at its best speed, and runs the slower RAM at its best speed.  Or, the motherboard might run both sticks at the slower "compatible" speed.

9. On NewEgg board, GSkill has commented that the XMP profile needs to be set to achieve the 1200/2400 speed. 

Good to know.

10 Clearly, I think that should be prominently mentioned in the Specs and not buried in a comment response.

It's mentioned on the GSkill web-page that I cited, for that part-number, namely "2133" and "2400" as different attributes.  They are not "hiding" the information -- it just requires significant knowledge of the two different attributes.

11. And in real terms... does any of this really matter? 

Yes.  2133/2400 = 88.875 %

12. Does the 2133 speed actually manifest itself in any way perceptible to the end user?? 

Would you notice a 11.125% decrease in your hourly pay? 

13. I.E. is it perceptible at all to the user in most typical PC uses like Moderate gaming, Internet videos, Word/Excel/Etc. (not large-scale data crunching) ?? 

Note that "everything" goes through the RAM, e.g., disk I/O, virus-scanning a file, downloading a Windows Update, and so on.

Note that real-time rendering of a moving avatar is "large-scale data crunching", requiring "gaming-level" hardware.

[SemiTechie]

Thanks for the reply and details.  Although GSkill may not be 'Hiding' the info, I believe most would assume that buying RAM advertised as 2400 speed would actually run at 2400 speed when installed under most conditions. The original HP installed RAM is running at 2400 and thus I bought RAM advertised as 2400 speed and assumed that installing 2400 speed RAM would run at that speed (as I believe most would); disapointed that is not the case here with this GSkill RAM.

May try a different Manufacturer to see if I can get a 2400 speed RAM  that will actually run at 2400 and will update this board if I find/discover any new info.

Glad to find this board as a great resource for this info!   

[SemiTechie]

I replaced the Gskill RAM that was rated at 2400 speed (but showing 2133) with a Crucial Balistix stick.  That stick showed the speed running at 2400, but showed a higher latency. 

Trying to determine if any of this matters... I ran a few benchmarks thru NovaBench and UserBench, only to find out that there really wasn't ANY significant difference between the Crucial at 2400 and the Gskill at 2133 (both running bench results within the same range for both Memory and total scores over 6 runs ea).

So it looks like 1 of 2 things may be happening here as I understand it.  

1. The Gskill is actually running at the 2400 speed when needed despite what Speecy is showing.

2. Or, the difference in the RAM running at 2133 Vs 2400 is virtually imperceptable, even in benchmarking.

One other intereesting note is that the Crucial RAM also showed the XMP profile to be active, just like the Gskill did, but the Gskill RAM only showed speed at 2133.

I'm guessing other more knowlegable folks here can offer a better interpretation of what I found, but I thought Id' share what I found.

[mdklassen]

This is quite a "follow-on" discussion to a post with an "Accepted Solution".

Next time, please open your own question.

> I ran a few benchmarks thru NovaBench and UserBench ...

> So it looks like 1 of 2 things may be happening here as I understand it.  

> 1. The Gskill is actually running at the 2400 speed when needed despite what Speecy is showing.

> 2. Or, the difference in the RAM running at 2133 Vs 2400 is virtually imperceptable, even in benchmarking.

2133/2400 --> 88.875 %

That should be a noticeable difference.

I would download a bootable memory-test utility from: www.memtest.org

and run it, one stick at a time, and compare the elapsed time to run one complete "cycle" of all the tests.

Tell us the results.

[hnelson7275]

I just updated the bios today and I can confirm that the speed of the original ram ran at 2400 ( I took a screen shot)  Now the same ram is clocking at 2133 after the bios update. Original Bios date was 3/30/2017.  It is obviously an issue with hp and the bios update.  They should fix it.