Better Together: Building a Strong Arrow Lake Ecosystem | Talking Tech | Intel Technology
- Better Together, it's a program that we put together working with our close partners on three main vectors. Motherboard, thermal, and memory. Of course, those three vectors, we are trying to best optimize the different systems that the ODMs and the motherboard providers are bringing by fine-tuning the software and selecting the best peripheral and hardware to get to the best possible performance. First, we are selecting the partners that best fit for the tasks. Once we select our partners, we are creating a workshop that we bring them together.
Each partner, including Intel itself, bring the added value to the point that we best optimize the performance. Altogether, it's bringing the different partners together to deliver a better-performing system. - The motherboard vendors actually come here to the overclocking lab, and we optimize with them their board for Arrow Lake.
They are really impressive in this area. They have a lot of tricks, but we join them with our engineers. We come up with some great recipes in the area of memory and beyond that really help ensure success. Then, we also work with the memory vendors. We have what's called the Extreme Memory Profile Technology, the XMP specification, and some self-certification guidelines we have. We work with those memory vendors to then certify their memory with our processor on a specific motherboard, and we put those up on the intel.com/overclocking website
so people can go and see which modules have been self-certified to work with this processor and this board. It's a really great program. It's a great way for consumers to understand that they're getting memory that's solid and will overclock as expected. - One of the things that's been a lot of fun is I've been doing this for quite a while, and the industry has really matured in the time that I've been here. Things that we used to have to do internally 100% of the way in order to get them 90% of the way there, we're able to now hand them the information. A lot of our ecosystem partners now have actual computational fluid dynamics capabilities themselves.
Instead of us needing to do all of that and then hand them just very simplified design guidance, we can give them the information that they need to run the models themselves and do the fine-tunings that they wanna do specific to their manufacturing capabilities or the type of industrial design they really wanna enable or some of the features that they wanna do. - So design is phases, but if we're talking about defining the product and design for specific feature, this is certainly something that we have a process that we are collecting feedback from our partners and feeding it into the definition of the product, and that's how we feed into the design. - We've been experimenting with releasing more detailed information to the ecosystem, and in fact, recently we had a symposium in Taiwan where we were able to provide more detailed information than we have in the past. And this allows our partners to use their new capabilities to really optimize their design specifically for the types of the market that they're wanting to go engage with and really optimize around.
- Before, I've actually never seen any roadmap from Intel. I was not able to see because we are not partnership. And then this year, we first time joined this program, and then we can see the roadmap and then your plan, and then we can prepare. - In order to develop this, it's a cooperation between Intel, us, as well as all the major motherboard vendors. We work with Intel's processor, and then we develop the high-speed memory.
And we also work with motherboard vendors to make sure that the BIOS tuning is up to par and all those three factors coming in together to make sure that this new Arrow Lake platform is performing where it's supposed to. - The whole idea of the Better Together, we are creating a workshop. So it's not just that they have remote access to those people. So we bring experts on site and working together with our partners to make it better on site. So it's all kind of in a face-to-face environment that you work together and come up with a solution at the end of it. (upbeat music) So we have four key partners that we are working on the motherboard as part of the Better Together.
We are working with ASRock, we are working with the ASUS, MSI, and Gigabyte. - MSI and Intel have been working for a number of years through a lot of different projects. And with the Better Together program, it started off in 2022. And obviously for Intel and MSI, we wanted to bring together the ultimate performance, the ultimate overclockability for all of our users, regardless if it's just the normal DIY users or the enthusiast levels. So we'll just, we'll be able to cover everything. - ASRock collaborate with Intel since from 2002.
And we did like participate a lot of OC workshops and also some OC seminar in Taiwan. And we also doing a lot of QDRs and QBRs every week to maintain the relationship between ASRock and Intel. - So a big moment for us came in 2006 when ASUS started ROG, Republic of Gamers. And that's our way of pushing for constant breakthroughs in both ASUS and Intel products.
And ROG lets us combine what we are both good at and we keep making great things together. - Intel and Gigabyte's partnership has continued for over three decades, 30 years. So we're deeply grateful for the staff and collaboration support from Intel. - So sometimes there are situations that indeed we are able to see better performance on motherboard systems.
And that comes with a specific implementation either from a package that we need to carry on the RVP for all kinds of debug logs that create certain level of penalty. So there are certain things that impact the performance or influence the performance in a certain way that on actual production motherboard, we have the flexibility to take advantage over where it's not very applicable for the RVP that we do internally. - For this generation, Z890 motherboards, we have a completely revamped our BIOS. So for each of our segments, we have different themes and different overclocking abilities throughout the integration of our BIOS.
- This year, Intel has introduced the RL ILM on the Intel Core Ultra processors. That's to better equip the motherboards for outstanding overclocking performance. - So the RL ILM was our answer to saying, okay, for those who want it, here's an Intel validated solution that keeps things flatter. But there's always a trade-off with everything, right? And so the RL is reduced load.
And in order to not have the IHS bend as much as it is, you reduce the load on it. But the problem is reducing the load means that you don't have as much of an engagement on the pins in the socket itself. If you do that, you still need to have those pins making good contact with the bottom of the substrate. And so now the heat sink itself has to put more load onto the IHS in order to make sure that the socket is fully engaged. Normally with the ILM, when you engage that lever, the reason why you're doing that is 'cause there's all of those pins underneath in the socket that make contact with all the dots on the bottom of the substrate, right? Each of those pins is a spring.
And you wanna make sure that you're getting contact with every single one of them. So that means you have to push every single spring down. And when you have that many springs, it's a lot of force in order to compress everything, right? And so that was the job of the ILM is to make sure that you're providing enough load that whether you install the heat sink correctly or not, electrically, the part will still function. - ILM is like the solutions offered by Intel officially. So people don't need to modify the aftermarket, the CPU frame, yeah, to make the CPU cooling better because after we have the ILM, I think it's improved a little bit. Yeah, make the surface like flat.
Yeah, so you can get better the heat performance. Yeah, after that. - For sure, for all Z890, we have the RL-ILM implemented on board. So then also we have provided our own cooling solution for the AORUS Waterfront II 360 AIO that also can fully support the RL-ILM, which is the Z890 motherboards. - For this generation of motherboard, we've put in a new RL-ILM retention plate, which that does is that it helps with the CPU coverage on the socket with the CPU and for our waterblock cooling solution to have much more of a tight gap.
- We use the Intel that suggests the default setting for our Z890 and Arrow Lake platforms. So all the boards will follow the Intel default settings to guarantee the best performance. It's tuned by default.
Yeah, so you don't need to choose any profile. So when you just pull up and it will be like Intel default settings. - We can use the Intel default settings. We don't see any impact or like a drawback on the performance. - So out of the box, we'll be running Intel's default BIOS settings.
So all of the PL1, PL2s are all within the guidelines. - For the DLVR change, we know that this is a very unique or like exclusive feature that is first introduced on the Arrow Lake CPU. We have forged a very, very strong cooperation together and try to make sure that we have enough BIOS supports on that because after all, it's come from the CPU logic directly. We also co-work with our software partners, informing them that we're going to have this change.
And they're also cooperating with us for a lot of like checking the address, how do you read the DLVR because the vCore itself and also how do you change. - So for DLVR, this is a settings that will be, we pretty much follow Intel's guidance and we'll have it in our BIOS to have it as on for most of our users, but obviously for some of our hardcore LN2 overclockers, they will still be able to access that feature and be able to turn it off. - I think that's quite good design. Yeah, because the power efficiency is pretty good.
And also I think more accurate and can make the CPU like work at better temperature and frequency, I guess. (upbeat music) - You know, we've had this strong relationship with memory vendors going back to the introduction of XMP and those relationships have only strengthened. We work with them very carefully as we introduce each new platform. Arrow Lake is no exception.
Working on new memory technology this time around and reaching new speed levels that really the industry has never seen before. - We started this year with Intel through the Better Together program. We offer the memory to Intel and then they help us to optimize our performance. And if there is any error or something, they also let us know in advance before launching. - ADATA is extremely pleased to be part of the Better Together program with all the partners that work together on making sure that everything is validated and everything works well as a one. - Kingston has been an Intel partner for over 25 years and then recognized by Intel in global memory channel and in supporting Intel platform with high quality and Intel validate the memory solution.
- Since 2023, we cooperate with the CERA team. We should be called CCG team. So from that time, as you know, module house actually work with the motherboard vendor like ASUS, MSI, these guys.
So I think it's very happy to work with them with Intel also. - We've been working with Intel since before DDR3 days. And so we were on board with the first XMP program with Intel and up until DDR5 with Intel XMP 3.0. And we were recently working with Intel on the Arrow Lake platform. And that's what's allowing us to improve upon system performance as well as memory speed performance on the latest gen platform. - In 2008, when Intel decided to launch the extended memory profiles, ADATA decided then to start XPG or Extreme Performance Gear, which is our higher end gaming brand.
And obviously we are very, very excited now to be part of the Better Together program. - Due to the great partnership with Intel, we have early access to the reference platform and processor. That helps Kingston to verify our memory design compliance Intel's architecture.
And then Kingston also work with leading motherboard and system manufacturer. And then we exchange the hardware and then work closely together to cross-qualify our products to provide a trouble-free experience for our mutual customers. - With Arrow Lake, we have redesigned our memory reference code. We're actually planning internally to achieve certain levels of MRC capability. So that's memory reference code. And that's what we would hand off to our motherboard vendors.
So our motherboard vendors would then have this base MRC that is essentially more capable than the past. So the team within Intel, the memory reference code team has put a significant amount of effort into essentially improving the ability to support these higher speeds. So MRC is really important because its job is to essentially train. There's a training method where these frequencies and latency settings are actually brought up early on in post. And so that memory reference code really had to comprehend the capabilities of the Arrow Lake memory controller and really to take advantage of that.
So we're excited with the quality of that memory reference code. Well, this is really exciting because our motherboard partners take our base memory reference code and they will then further tune it to match their board and also maybe their own overclocking goals for memory. And so it's amazing to see what they can do. They can take something that's starting here and bring it up to that next level. This MRC collaboration is really iterative.
So it progresses over time as we'll raise that baseline up a notch and we'll bring it up another notch and they'll keep updating and taking advantage of each step forward. So you could imagine as the project progresses and we're reaching higher and higher frequencies, the MRC will be adjusted. - We actually work with major top motherboard manufacturers like Asus, MSI, Gigabyte, ASRock. They are supporting us for all the day. We always send the memory to them.
They also let us test their motherboard. - So we get access to early prototypes. We get to do a lot of validation and run an extensive suite of tests during the engineering sample phase. - We get the motherboard first, our platform, and then we test it and then it cannot be perfect.
So we ask them to keep changing the BIOS and then they give us BIOS every day for last few months to upgrade and then we work together to make a highest speed megahertz. - So basically this is our way to cooperate with them because for the DDR modules is something like client component, not like a motherboard or CPU. So we have to follow CPU spec and the motherboard BIOS to check everything. - So most recently we've been working together with Intel under the Better Together program and there's been a focus with Arrow Lake platform to push CU-DIMMs and this is a new technology we're really looking forward to 'cause that helps push the memory speed performance up to a whole new level.
- A CU-DIMM is very similar to the traditional UDIMM but the main difference is that there is a redriver, what we call a clock driver internal to the memory module and so it will generate essentially its own clock nice and clean and that can enable higher speeds especially for those kind of mainstream and low frequency XMP levels and regardless of some of the characteristics of the motherboard that could be an impediment. So like even lower price point boards could actually really benefit from this. - One of the unique selling points of CU-DIMMs is the fact that here in the bottom of the center of the board actually you have a clock driver or CKD chip that essentially what it does is provides timing separately from the original way to provide timing which is pushed from the CPU into the bus and then to the memory module.
These are really the key secret sauce that allows these type of modules to be more stable and have higher frequencies 'cause at that point you don't rely on a clock that is far away, it's really on the source directly on the board. And that also, even though it's a cost other from a chip perspective, it actually makes the whole module cheaper to produce because there is less connections. - This CU-DIMM, the new memory technology actually enhance the signal integrity of memory.
It can bring more stable experience to the users. - So while the CU-DIMM is amazing and it's important for the majority of customers, some of our XMP customers may choose to either buy modules with or without CU-DIMM and essentially bypass it to reach those incredibly high speeds. So if you're really going after, say well over, say the 8,400 or 9,000, you'll probably end up bypassing that clock driver. But the cool thing is you have choice.
So if you want stable, low price points, you have that CU-DIMM. And if you wanna go for those high, high speeds, you can also disable it. - With Intel platform, obviously it's with Intel XMP 3.0. And this is continued to be a very easy way for end users to overclock, whether it's UDIMM or CUDIMM on their platforms. And we focus on making the memory kit as easy to use as possible. So if you have the right motherboard, right CPU, update the BIOS, and then you plug in the memory kit, enable XMP, and we've done validations with Intel under the XMP program.
And that would just work at the rated speed. As far as extreme overclocking goes, we've always been working with extreme overclocking. We're hoping this Arrow Lake platform can, with its new IMC, with its redesigned, compared to previous generation, that we'll be able to hit new heights in terms of extreme overclocking. And this is something to look forward to as we get to the launch of Arrow Lake.
- We love working with memory vendors throughout the industry. The overclocking memory speeds you will see are a result of that great, strong partnership with the whole ecosystem. - So every generation, we release a lot of documentation to help our partners understand what are the changes from the previous generation, where are the hotspots located, how much heat do they need to dissipate for each of the different market segments.
This time around, we've done things a little bit differently. There was a symposium that we had with the ecosystem where we introduced a little bit more collateral than we normally do. And when I say collateral, I mean the detailed information of what's inside the CPU before launch so that they're able to go in and optimize their designs so there's a little bit more detail that we release this time around so that way they can position themselves just a little bit better for Arrow Lake S and really help it shine in terms of making sure they're providing really good cooling capabilities for our parts. - In 2018, we expanded our business to include all-in-one liquid coolers to provide more comprehensive options for our users.
This program has given us early access to thermal profiles so we can develop optimized cooling solutions for Intel's latest processors. - During development process, we will receive some information from Intel like a thermal design guide. It's very important to us for CPU cooler design. - This information is very important for us to design a new liquid cooler.
If we don't understand, we don't know this information, we cannot do some improvement in the liquid cooler design. - That's our challenge all the time is I'm putting all the power into a smaller area every generation and I still need to conduct it out and get it into something that we can cool with the thermal solutions that we have available in the system. So power density is increasing each generation.
I still have to keep it within roughly the same temperature limits. One of the things that we did with Arrow Lake is we started interleaving the P cores and the E cores to help spread that heat out a lot. And this has two benefits. One of them, when all the cores are active, it keeps all of the hotspots a little further apart. But then also when you're doing things where you just need all those P cores operating, but maybe not so much the E cores, say for example, like gaming, for example, right? Your eight P cores come up to speed and the E cores are at low power and that helps spread the heat out quite a bit.
So our previous generations, all of the P cores were lumped together on one side and all the E cores were lumped together on another side. But we've spread the heat out quite a bit by moving those around. So it was a noticeable improvement for us and we're pretty excited about that. - So during the Better Together program, Intel delivered this message in early, so we have enough time to design our new liquid cooler. - Intel provided us with things like some early CPU samples and thermal guidelines to identify some potential issues and validate thermal performance. - We received a lot of information from the Intel and this information is very important.
For example, Intel let us know they will change the ILM because the original Raptor Lake refresh the ILM maybe have some little improved when they change to the Arrow Lake. Second information is the Intel provide the P core and the E core. They tell us the thermal distribution is different. So maybe they have some improvement so the thermal hotspot will move from the center to the up position. - So Intel's information and some design guide is very important to us and update to date closely every day. - By working with Intel and Asetek, we optimize the co-play and the pump design to enhance cooling efficiency.
- We have three key features optimized for Arrow Lake CPU. Like first is to optimize the mounting pressure and the second, it's larger copper base design. And the third part, it's reverse floor water design. The thermal solution, it's a key element. So you can go to thermal solution that will cool down the part in such a way that you can overclock to much higher than the nominal frequency.
(upbeat music) When working with partners, there are collaterals that we are providing that include the electrical, thermal, mechanical and overall design guidance. That's the fundamental guidance of how to put the system together. We are bringing our capabilities for tuning along with the capability comes from the different partners on the memory side, on the thermal solution side as obviously the motherboard providers.
And together we are bringing the most optimal solution to the market. (upbeat music) (soft music)
2024-10-18 05:23
