The MSI GS66 gaming laptop sells with DDR4-2666 memory despite the new Intel 10th gen platform supporting DDR4-2933, so let’s find out what sort of a performance difference this actually makes in games and find out if it’s worth upgrading to faster memory.
By default, the DDR4-3200 memory runs at DDR4-2933 at CL21 while the DDR4-2666 ran at CL19, however I was able to get my 2933 running at CL19 too and match all the timings.So we should be in for a pretty fair comparison in terms of the memory speeds. Both configurations were 2 16gb sticks in dual channel.
You can tune the memory in the BIOS, just press
Delete during boot up, then once in the BIOS press Right Shift + Right Control + Left Alt + F2 to enable the advanced BIOS mode. From here go to advanced settings, down to the overclocking performance menu, enable the overclocking feature, and go down to the memory overclocking menu down the bottom. Within here you’ll have all the memory options you could want. You don’t have to tune the memory, but by default most DDR4-3200 kits will run with slower timings than the 2666 sticks MSI are using.
With that in mind, let’s get into the gaming tests. Battlefield 5 was tested running through the same part of the game in campaign mode. I’ve got the stock DDR4-2666 results shown by the purple bars, and the faster 2933 results in the red bars. There was a small improvement seen to the average frame rates, however there was a much larger boost to the 1% low performance comparatively, though this was one of the better results out of the titles tested. Call of Duty Modern Warfare was also tested in campaign mode by running through the same section of the game. There was only a slight edge to the 2933 configuration this time, though at the end of the day it’s extremely negligible in this one. Shadow of the Tomb Raider was tested using the games built in benchmark tool, and this was one of the few games tested that actually saw slightly worse results with the faster memory. I’m not too sure why that was the case, but either way the difference is only a few frames. PUBG was tested using the replay feature with the exact same replay file used in all instances. There was an above average increase to performance with the 2933 memory in this one, at max settings the average FPS was 5% faster, however check out the differences to the 1% lows. There’s a massive 58% boost to 1% low performance at very low settings with the faster memory, though this was the highest change out of all games tested. Fortnite was also tested using the replay feature with the same replay file in all cases. There was a much smaller difference observed here comparatively, though there was still a slight edge to the 2933 configuration. CS:GO was tested with the Ulletical FPS benchmark, and there were minor differences at max settings, the 2933 config was just 1.3% faster in average FPS, however with everything at minimum it was now 7% faster. Dota 2 was tested playing in the middle lane, and was another game that despite seeing an increase in all instances with the faster memory, it was only subtle and nothing that impressive, still better than a loss I suppose. Speaking of a loss, control was seeing lower average FPS at all setting levels with the faster memory, however the margins were extremely minor, margin of error stuff realistically, I’d say it’s performing about the same regardless of memory at all settings. Borderlands 3 was tested with the games benchmark tool. The average FPS is basically the same with either memory option between high and max settings, however the faster memory pulled out ahead between medium and very low, although even best case at minimum the 2933 memory was just 4% higher in average FPS and six and a half percent higher for the 1% low. Overwatch was tested in the practice range, so performance is better than actual gameplay, but it allows me to more accurately perform the test run, ideal for a comparison like this, either way basically no real differences here. Far Cry New Dawn was tested with the games benchmark tool, and there was an improvement seen at all setting levels with the faster memory, though not by too much. I’ve noted this game as being more CPU sensitive in the past, and memory is definitely tied into that. The Witcher 3 also saw some improvements with the faster memory, however once more like most of the other games tested it wasn’t that much and wouldn’t realistically offer a perceivable difference in most cases. Assassin’s Creed Odyssey was tested with the games benchmark, and the performance varied based on the setting level, but again not that much difference regardless of setting level, some saw a slight increase while others were down.Rainbow Six Siege was also tested with the games benchmark tool, and although the average FPS was ahead slightly for the most part with the faster memory, interestingly the 1% lows were down at all setting levels, I’m not too sure what the deal was here. Ghost Recon Breakpoint was tested with its benchmark tool as well, average FPS was quite similar with a larger boost seen with the faster memory at max settings, though it was still only a two FPS change.
These are the differences to average FPS over all 15 games tested with the highest setting preset. On average the DDR4-2933 configuration was just 1.2% faster than the DDR4-2666 configuration that the GS66 ships with, so not that big of a difference. There were some titles that saw basically no change, or even slightly worse performance, though 1% is pretty negligible and probably margin of error range anyway.
When we look at the differences to average FPS at the lowest setting levels there’s a slightly larger 1.5% improvement to average FPS with the DDR4-2933 configuration. Presumably the memory matters more as we’re more CPU bound, and some competitive games like PUBG and CS:GO saw a fairly respectable 7% performance improvement.
There are much larger swings when we look at the 1% low performance at minimum settings. There’s the strange Rainbow Six Siege result down the bottom, not sure what the deal is there but it’s the only game I tested with Vulkan. Meanwhile, PUBG and Battlefield 5 saw huge boosts to the 1% lows with the faster memory, so 2933 should offer a much more stable experience there.
I’m testing the GS66 with the Intel iGPU disabled, it may be possible that the higher memory speeds are more beneficial when the Intel graphics is enabled, however I haven’t tested that here as this review was meant to be specific to the GS66, and it’s my honest opinion that most people playing games will be using it in discrete mode.
Outside of gaming, I’ve also run the AIDA64 memory benchmark The DDR4-2933 configuration was hitting 9% faster read speeds here, so a far higher result than what we saw in the games, but this would be because the bulk of the games aren’t constrained by memory bandwidth.
So in some games, the memory change can make a huge difference, but once we average multiple titles together, on average it’s nothing that impressive. Unless you’re playing a game that sees a good boost with DDR4-2933, it’s probably not worth replacing the 2666 sticks that ship with the GS66 with faster ones.
At the same time though, the GS66 seems to be positioned as a more premium gaming laptop, so it seems a little strange to me that they didn’t just use the best possible memory configuration for it, maybe they just have a ton of 2666 sticks they need to use up or something. I’d be interested to hear your thoughts on the performance differences, and whether or not MSI should have just given the GS66 faster memory out of the box, let me know down in the comments.