MSI X299 Breaks DDR4 Memory World Record With 5500-DDR4 Speeds

[Raevenlord]

Toppc Pushes the X299 Gaming Pro Carbon AC Motherboard to Reach New DDR4 Frequency Milestone
One year ago, MSI's in-house overclocker, Toppc, broke the DDR4 5GHz barrier using G.SKILL memory and the MSI Z170I GAMING PRO AC motherboard. Today, 1 year later, he raises the bar once again and became the first ever to push DDR4 speeds to 5.5GHz under liquid nitrogen cooling using the new MSI X299 GAMING PRO CARBON AC motherboard. With the MSI X299 GAMING PRO CARBON AC, MSI's most customizable high-performance Intel X299 based motherboard, Toppc was running G.SKILL DDR4 memory with an unbelievable DDR4-SDRAM clock speed of 5500MHz. This world record shows MSI's dominant position on performance for X299 by using MSI's unique and patented DDR4 Boost technology. The MSI X299 GAMING PRO CARBON AC is not only feature packed for gamers and great for case modding, it is also perfectly suited to deliver power for heavy gaming & overclocking sessions.





X299 Gaming Pro Carbon AC: Perform in Style
The world's first motherboard with fully exchangeable heatsink covers and X-Mounting accessories to support 3D printing and modding, combined with the latest generation MYSTIC LIGHT RGB LEDs, makes the X299 GAMING PRO CARBON AC any modders' dream!




Just activate the one-second Game boost function and performance raises to the next level like manual overclocking, without breaking a sweat. Furthermore, with the addition of 2x M.2 Shield, any fast PCI-Express SSD is sure to deliver maximum performance. With Turbo M.2 and Turbo U.2, simply connect any of the latest generation NVMe based SSDs with ease, providing you a lighting fast storage transfer experience. With unlimited customization options and packed with exclusive gaming features, the renowned GAMING PRO CARBON is ready to astonish once again in any gaming build

View at TechPowerUp Main Site

 

[Nuckles56]

@Raevenlord you were beaten to it https://www.techpowerup.com/234146/g-skill-memory-breaks-ddr4-5-5-ghz-world-record-barrier

 

[ne6togadno]

uhm
https://www.techpowerup.com/234146/g-skill-memory-breaks-ddr4-5-5-ghz-world-record-barrier

 

[Raevenlord]

@Raevenlord you were beaten to it https://www.techpowerup.com/234146/g-skill-memory-breaks-ddr4-5-5-ghz-world-record-barrier

Aye, but this is MSI's PR. Bet they want their name and logo on the front page as well about that

 

[-The_Mask-]

Aye, but this is MSI's PR. Bet they want their name and logo on the front page as well about that

But it's still only 2750MHz...

 

[dj-electric]

But it's still only 2750MHz...

No... not this $hit again. Please.

 

[-The_Mask-]

No... not this $hit again. Please.

Totally agree with you, so stupid that people call it 5500 MHz, while HWbot and CPU-z both clearly state that it's 2750 MHz. It's like we are going back 15 years were people didn't understand the working of DDR memory.

 

[MrGenius]

No. It's not stupid. That's what people call it because that's what they've always called it. Always have, always will. And they have good reason to. Since DDR is double-pumped. Effectively doubling the data rate. Making it(the data rate of DDR RAM) EXACTLY equal to the data rate of single-pumped RAM @ x2 the clock rate MHz. Hence it being totally accurate to state the speed of DDR as x2 the clock rate MHz. Since that's precisely how fast it is working. Wonder why your DDR sticks are rated for whatever MHz they're rated for, and NOT rated for the clock speed MHz you run them at? That's why.

We're not going back 15 years to any period of time when people didn't understand the working of DDR memory. Nothing has changed in the last 15 years in regards to people calling the speed of DDR as the effective data rate. Or rather double that of the given clock speed in MHz. That's the way that's always been...and always will be. I don't know what cave you've been hiding in to not have noticed it. But you might as well come on out and get used to it. Things aren't likely going to conform to your warped sense of reality any time soon.

 

[-The_Mask-]

Someone needs to learn the difference between MHz and MT/s... How can people not know the difference after more then 15 years?

 

[MrGenius]

Look...I understand what you're saying. And I'm not even saying you're wrong. What I AM saying is we're both right. MT/s or MHz...they both convey the same meaning. And people like you claiming to be "technically" correct by saying MT/s can do so until you're blue in the face. Because those of us that don't...don't care if you are "technically" correct. Why? Because you're splitting hairs. It's one number used to convey effectively twice the clock speed MHz(since 2 transfers occur during each cycle/Hz), or twice the number of transfers per cycle/Hz/second at a clock speed MHz. And, since that number is the same no matter what you call it, or how you calculate it, it doesn't really matter what you call it. It means the same thing either way. DDR is twice as fast, in effective MHz, or MT/s, as SDR. Twice as fast is still twice as fast no matter what you call it. So go ahead and feel intellectually superior for calling it whatever you want to call it. That's fine. But we're still going to call it whatever we want to call it too. Sorry.

 

[hapkiman]

Someone needs to learn the difference between MHz and MT/s... How can people not know the difference after more then 15 years?

What about the difference between then and than?

 

[R-T-B]

Someone needs to learn the difference between MHz and MT/s... How can people not know the difference after more then 15 years?

Hz means "times per second." It says nothing about wave forms. If it's double pumped on a wave form, it stands to reason that it's still doing twice as many transfers per second, even if the actual wave form cycles at half.

Your point is silly at best.

 

[-The_Mask-]

No it's not. For example, what are you gonna do with DDR5? Some people already call 2000MHz GDDR5, 4000MHz GDDR5. That's silly. Because it's 2000MHz or 8Gbps, calling it 4000MHz doesn't make any sense. Same problem are you gonna have with DDR5, some will call it 2000MHz, some 4000MHz and others 8000MHz. And what if you actually clocked your DDR5 at 1000MHz, some say 2000MHz and some are gonna say 4000MHz. Because some people are making up numbers it's extremely confusing. If someone is talking about DDR5 at 2000MHz it than also can be 1000MHz, 4000MT/s or even 8000MT/s. Just because some people can't get their facts right.

 

[Toothless]

Man this thread megahertz my head.

 

[Aquinus]

Hz means "times per second." It says nothing about wave forms. If it's double pumped on a wave form, it stands to reason that it's still doing twice as many transfers per second, even if the actual wave form cycles at half.

Half a cycle is still half a cycle. When you talk about something running at a frequency, in particular hertz, 1Hz is literally one full cycle, by definition. Each rising and falling edge is a beat, not a cycle. So it's technically correct to say 2.75Ghz and not 5.5Ghz because "transfers", "beats", and "cycles" are all different units describing something different and that saying 2.75Ghz and 5.5GT/s are describing two different things about the same circuit but, to call each rising or falling edge a cycle, is simply incorrect and misleading.

 

[Disparia]

Since this is still going on I'll just throw in that no, I don't believe this has always been the way. Honestly, to me it feels like a very recent trend to append Mhz to the name. Perhaps I just haven't been paying enough attention.

All I can theorize is that when DDR was new, there were many in-depth articles on the subject (as well as more trust-worthy websites for a techie to frequent) so we all used a stricter terminology. Subsequent DDR2, 3, and 4 releases focused on how they improved over the previous generation and less on the core technology so maybe it has blurred over time.

I personally like to use the proper naming, as there is less ambiguity when following the actual standard set by JEDEC and it results in 3 less keystrokes But, I won't call anyone out for this, the forums are casual conversation after all.

 

[R-T-B]

Half a cycle is still half a cycle. When you talk about something running at a frequency, in particular hertz, 1Hz is literally one full cycle, by definition. Each rising and falling edge is a beat, not a cycle. So it's technically correct to say 2.75Ghz and not 5.5Ghz because "transfers", "beats", and "cycles" are all different units describing something different and that saying 2.75Ghz and 5.5GT/s are describing two different things about the same circuit but, to call each rising or falling edge a cycle, is simply incorrect and misleading.

I'll concede because that actually makes decent sense.

 

[Aquinus]

I'll concede because that actually makes decent sense.

To add further clarity, what people a lot of time don't realize is that only the data lines are DDR in modern DRAM, control and address lines are SDR. So, those timings you configure for DRAM, those timings are literally measured in full cycles, not half cycles. ...to further hit the point home.

 

[simlariver]

Intel is just fishing for some good press on the X299 platform at this point. When did pro OC'ers had early access to hardware like this ?

 

[EarthDog]

Each AIB has them in house...so, they get it with the aib.

 

[cadaveca]

When did pro OC'ers had early access to hardware like this ?

most pro OC'ers work for one OEM or another and have had this level of access for well over a decade at this point.

 

[Hardware Geek]

Someone needs to learn the difference between then and than... How can people not know the difference after more *than* 15 years?

Someone needs to learn the difference between MHz and MT/s... How can people not know the difference after more then 15 years?