Kingston HyperX Memory Review Discussion

I run 2 virtual desktops and 3 virtual servers on my notebook

 

So, regardless of issues, are you guys seeing better performance at 1866 vs standard 1333? Better QuickSync performance on video renders?

 

No difference in real life performance since the memory downclocks to 1333MHz on dual core units and 1600MHz on quad core ones, despite supporting up to 1866MHz. Though realistically, even if it did, the benefits would be pretty small (ie. only visible on benchmarks) as increased frequency is balanced by increased latency.

 

Not true.

Have you tested it?

HyperX Memory runs on 930.5 x 2=1,861 MHz on my Thinkpad W520.

System feels more "snappy" when multitasking and changing windows.

 

Also check out this review:

The performance numbers from upgrading the memory was insane. We saw performance gains from 6% to 86% depending on the benchmarks with the largest performance gains coming from graphics intensive games

Kingston HyperX 8GB 1866MHz PnP Memory Kit For Notebooks - Final Thoughts and Conclusions - Legit Reviews

(unfortunately they made the same mistake as Notebookreview, comparing 6GB of slow memory with 8GB of fast memory)

 

I would call it the same MISTAKE! How can anyone compare smth with multiple differences??

 

Haven't compared this set exactly but I have tested similar sets of 8GB DDR3-1066 to 8GB DDR3-1600 and have not found any noticeable benefit. Then again this is with a SSD and discrete GPU. Don't have a 2nd generation Core i series to test yet.

 

Update: Kingston are sending me two full kits to test, 1600 and 1866. Being a Lenovo Premium Business Partner I have aligned 3 different systems for that test, a 2720QM, 2820QM and 2920XM. I will test all systems in cominations of 8Gb and 16gb, at 1600Mhz and at 1866Mhz.

This should clear everything up I hope.

(Incidentally, Kingston mentioned that 2x4Gb is supported, 4x4Gb is not, regardless of memory speed. Reason potential memorycontroller overload.)

 

Hmmm. 4x4GB HyperX works fine with my Thinkpad W520 and apparently with notebookreview's Thinkpad W520 in their tests as well.

Notebook Reviews - Laptop Reviews - Picture

 

That's what I said previously. I wish you good luck with tests. Looks like extreme version has not so stable memory controller because of high frequencies I guess.

 

For the first time, I have now been able to recreate the sleep/resmume issue with HyperX and the Thinkpad W520.

I just noticed that when I use "balanced" I have the sleep/resume issue with the 1866Mhz RAM, but not in "maximum performance".

Perhaps that's why the reviewers never encounteres it, because you would do a review in "maximum performance" mode.

 

Search around the forum, there was a thread talking about the different profiles impact on performance and battery life. I believe it was concluded that balance was the best but if you want the in depth analysis of why, you'd have to read that thread.

 

Received a 4x4Gb HyperX 1600Mhz kit today. None of the issues that I encountered with 1866Mhz have returned. The machine performs flawlessly. No hickups, solid performance, resume from standby works, power on by finger swipe works ...

 

Cool!
Did Kingston send you a fresh set of 1866MHz modules to test too?
- Did you test the 1600Mhz set using the "Balanced" power plan?
- Which W520 BIOS version do you run?

 

- Yes I should receive a new kit of 4x4Gb HyperX 1866Mhz next week or so
- I tested at Max Performance and at Power Source Optimized
- 1.25 (latest)

 

Thanks.

Could you also test the 1600Mhz kit using the " Balanced" Power Plan in the Windows-7 " Power Options" (click the battery/power-plug symbol in the bottom right windows 7 tray)

PS: W7 "Balanced" is how I managed re-create the sleep/wake problem with my 1866Mhz kit with BIOS 1.25 (latest)

 

In Balanced mode, with BIOS 1.25 1600Mhz works fine in all counts. Memtests also are ok.

I will receive the new 1866Mhz kit this afternoon and will re-test all scenario's. Update will follow.

 

Update: HyperX 1866Mhz kit installed and the first two quick tests with 4x4Gb=16Gb immediately failed. Lenovo toolbox indicated a memory error, followed by a BSOD.

I haven't started testing combinations of 2x4Gb yet but this already indicates that for the 2920XM the treshold tolerance definitely lies somewhere between 1600Mhz and 1866Mhz. The 1600Mhz modules work flawlessly in any given scenario.

 

Also try to put 2x4GB 1866 in different slots so you will check them in dual mode in paired slots and in single mode.

EDIT: Wditters could you test 1600 HyperX on 1-st Gen I-core?! Please! When you do post screenshots of CPU-z where I can see timings and latency (clocks). If you have Sony laptop (VAIO F for example) I would be happy

 

Thank you for the test. This is very interesting!

Since the 1600MHz HyperX memory modules have a better latency (9 vs. 11), they may be a better choice, even for the i7-2820 and i7-2720.

 

it doesn't matter. it has 11CL on 1866 but downgrading to 1600 it has 9CL.

 

Isn't 9CL better?

 

Meory 1866Mhz can work:
1866 - 11CL 11/1866*100= 0.5895
1600 - 9CL 9/1600*100= 0.5625
1333 - either 7 or 9 7/1333= 0.5251
1066 - either 5 or 7
lower don't know.
Lower nanoseconds is better.
So if I am right with timings schedule then working on processor which supports 1333 max will use RAM even faster than on 1600.
That is why I asked wditters to post screenshot of shedule with timings/latency. In any case RAM WILL DOWNGRADE TO 1600 cl9 BECAUSE 2-ND GEN i-CORE DO NOT WORK ON 1866. I saw one guy wrote his memory does not work on 1866 while someone wrote it's memory does work. Strange Memory.
Usual memory (not hyperx):
1333 = 9CL 0.675
1066 = CL7 0.657
Difference is negligible

 

Ahhh. I did not know that. Thank you for the explanation!

Do you have any explanation why 1866 HyperX would work fine with Lenovo W520 i7-2820 but not with wditter's W520 i7-2920?

Attached is the screenshot for my W520 i7-2820
it seems to work fine at 1866MHz with my W520

 

Welcome. I turned many I-net pages till I got that way to count performance.
Your sticks work at maximum as other guy wrote. He also posted his screenshots.

I have couple thoughts but wditters said that he is Lenovo business partner and he gonna check on 3 Lenovo laptops with different processors. Lets wait and see if it is processor make such an influence on RAM.

@Wditters. You still gonna use 3 laptops, don't you?

 

Thing is at 1333mhz you are pretty saturated so you are looking for latency reductions.

It's when running a higher powered integrated graphics solution that bandwidth becomes the leading factor again.

400 shader part supporting 2000mhz ram ftw.

 

My CPU-Z with 1866 HyperX on 2720QM looks identical.

It passed all Lenovo Toolbox Diagnostics tests and also Windows Memory diagnostics.

Still investigating stand-by issue on v1.22 BIOS.

- ngm

 

A review with different memory speeds (1600 - 1866 -2133):
http://aphnetworks.com/reviews/kingston_hyperx_khx1600c9d3x2k2_8gx_2x4gb/6
There does not seem to be much difference...

 

And here are some PCMarkVantage Is Your PC Fast Enough? PCMark Vantage Basic Tells You | PCWorld benchmarks:

W520 i7-2820 with 4x4GB 1333 Mhz RAM (500GB HDD): 8,614 PCMarks

W520 i7-2920 with 4x4GB 1333 Mhz RAM (500GB HDD): 9,970 PCMarks

W520 i7-2820 with 4x4GB 1333 Mhz RAM (single Intel 510 SSD): 16,302 PCMarks

W520 i7-2820 with 2x4GB 1333 Mhz RAM (2x240GB Intel 510 SSD in RAID0): 19,000 PCMarks

W520 i7-2820 with 4x4GB 1866 Mhz RAM (2x240GB Intel 510 SSD in RAID0): 19,228 PcMarks

i7-2820 vs i7-2920 improvement: +15%
HyperX improvement: +1%.
Sata III SSD improvement: +89%
RAID0 improvement: +16%)

Note: Per PCMarkVantage, the "processor clock" runs slower (3090 MHZ) with 4x4GB than with 2x4GB memory (3289 MHz). Does anyone know why?

 

Is it the consensus that the 1600Mhz HyperX RAM is the optimum performance/reliability solution for the Thinkpad W520 with Intel i7-2920XM processor?

 

Just added 8GB of the Kingston Hyperx 1866 to my Thinkpad X220 (i7-2620M). Prior to that I had the Lenovo stock 4GB.

My winsat mem speed went from about 9400 MB/s to about 17575 MB/s.

My Windows Experience Index for RAM went from 5.9 to 7.6; my scores for Aero Graphics) from 4.6 to 5.2, for "gaming graphics" from 6.1 to 6.2.

Ran the Lenovo Toolbox memory tests and memory stress test, all passed.

 

WEI very often shows 5.9 for 4GB of RAM on Windows x64. Try to puy in only 4GB of HyperX to see)
Speed increases thanks to higher frequency. but it doesn't help much only that couple %

 

Reading this thread I decided to share my experience with W520 (i7-2820QM) and Kingston HyperX 1866MHz modules:

1. tested each single one (after I got two "faulty" replaced by Kingston - 4 same type of modules tested) in slot 0 (1x4GB configuration) - numerous ThinkVantage mem tests, windows mem tests and memtest86+ - all OK;
2. tested in pairs in slot 0 and 1 (2x4GB) in same manner - all OK;
3. the moment i populate all slots (4x4GB of HyperX 1866MHz) I got BSODs immediately during boot, and when I run memtest86+ from USB stick then errors pop immediately on second test (time <30sec, where all test last 2 hours or so).

I have BIOS 1.26.

Kingston denied any responsibility - also for 1600MHz modules - just in case, since I asked for compatibility. There is relevant statement on their website.

My all 4 modules are same: rev. 9905428-048.A00LF, manufactured in week 20. You may have luck then - it works on some quad machines (16GB) as mentioned here earlier.

In 2x4GB (slot 0 and 1) 1866MHz config I have WEI of 7.9 for memory, and winsat mem command returns bandwith of abt 24200-24300 MB/s, which I guess is close to max for this machine. I never trust synthetic tests, and the only thing I can tell of difference between original 2x4GB 1333MHz Elpida - it is palpable. I work a lot developing pictures from RAW formats, so memory is a key factor for me. I aslo have Intel 510 SSD of 240GB cap as primary drive.

Question to all: could you please run winsat mem command to return bandwidth for following configurations (on both i7-2820QM, and i7-2920XM):
1. 4x4GB 1866MHz HyperX
2. 2x4GB 1600MHz HyperX
3. 4x4GB 1600MHz HyperX

I am siply very curious what the difference is in compare to my current configuration.

Any idea if BIOS update could possibly sort compatibility issue? Some other makers machines with same platform/processor are reported to run smooth with 4x4GB of 1866MHz modules.

As for sleep/resume issue: I have it in both high performance and power source optimized modes with 1866 modules, so I do use hibernate or shut-down only. Warm boot delay is gone when AMT is disabled in BIOS.

Many thanks in advance.

 

I wanted to know HyperX 1866 Mhz 8GB VS 16GB is that necessary? Will the 16GB perform better then 8GB?

 

Capacity is totally separate from speed.

 

Thanks bud , rep+

From what I heard if different ram will down clock it back to 1333Mhz , If 2GB + 2GB + 4GB + 4GB (total 12GB) @ 1866Mhz will it down clock the PnP ram?

 

If there is any 1333 RAM then it will downclock. Another scenario? Depends on memory controller (CPU) and brand (notebook, motherboard).

 

How about m18x , the rig in my signature. Usually people brought 2x4Gb(8GB) , 4x4GB(16GB) I wanted to get 2x2GB(4GB) 2x4GB(8GB) PnP is something new & a bit expansive , I would like to make sure before I pull the trigger , I'm too frustrated to wait 2133Mhz ram...been waiting almost half year..

 

Been a while, but I came across this article and the test methods are quite definitive.

AnandTech - Sandy Bridge Memory Scaling: Choosing the Best DDR3

 

This is what I have just wrote there:

 

any word if these are ivy bridge compatible?

 

any ddr3 is ivy bridge compatible. and i assume 1600 ram will work fully on any ivy. but you better buy cheaper corsair vengeance 1600 ram. it has no uselss metal shields.