Just recently, the SCYROX V8 left a lasting impression due to its near-unbeatable value proposition. Yet, on paper at least, the RAWM ES21M is almost a match: at the same $69.99, it comes with PixArt's latest PAW3950 sensor, 8000 Hz wireless polling, and a weight of just 47 g. While the V8 is even lighter at 40 g, the ES21M is a decent bit larger, and too does without holes or an open-bottom design. Aside from being larger, the ES21M also presents a different shape, with a centered hump and curvature closer to the Pro X Superlight, though without being a clone. In terms of build quality, the ES21M arguably fares even better than the V8. On my sample, there is no rattle when shaking, no creaking or flexing of the shell when applying lateral pressure, and actuating the side buttons by pressing below them is impossible no matter how much force is used. Looking inside the mouse, it becomes clear as to why the ES21M is this solid: a design not unlike the Razer Viper V3 Pro is used, where additional PCBs only holding the scroll wheel and main buttons as well as the side buttons are placed on a plastic assembly providing structural support. This comparison already serves to illustrate that in terms of engineering and internals, the ES21M punches above its price class.

This also extends to the switches used in the ES21M. For the main buttons, RAWM has opted for switches from Huano, and more specifically of the green transparent, white plunger variety. These aren't as common as the blue transparent shell, pink plunger variant, though in practice, they feel quite similar. Due to moderate pre and high post-travel, these switches don't feel quite as snappy as they could, and some lateral button movement can be provoked as well, although I didn't notice anything during actual use. The side buttons likewise come from Huano and are surface-mounted, which typically results in a less than ideal button response due to the smaller switch footprint. Not so on the ES21M, as pre and post-travel are minimal, the actuation point is even, and the button response itself pleasing. The scroll wheel encoder is a yellow/blue encoder from TTC, which is dustproof and provides rather good tactility at manageable noise levels. The feet are made of pure PTFE, and resemble the feet used on recent LAMZU or Incott releases, with the adhesive being smaller than the PTFE proper. Glide is good in any case, and the large footprint will be appreciated by Zowie or VAXEE users. Not so good is the coating, however. The main issue here is that the coating acts as a magnet for fingerprints, and actually getting them off can prove difficult to impossible. Essentially, once the ES21M has been touched, getting it back to looking pristine becomes but a wish.

In terms of software, RAWM is doing things a bit differently than other manufacturers. Most competitors simply opt for an off-the-shelf software solution, such as the one from CompX, whereas the software for the ES21M is self-developed by RAWM. Moreover, the same software is available either as a web driver, which connects through WebHID and therefore will only work on Chromium-based browsers, or as a regular local application requiring installation. Functionally, there is no difference, though the web driver isn't fully in English. The application manages to be light on resources, as the RAM footprint is small, and there aren't any additional processes permanently running in the background. In short, the choice is given to the user, and both the available options are competently realized. In terms of features, the software is more extensive than others. Aside from the usual options including sensor run mode, polling rate, MotionSync, ripple control, or angle snapping, the software also features several more unique settings. For instance, the glass mode of the PAW3950 may be forced—by default, the sensor enters said mode automatically based on surface properties, but with this toggle, the mode can be permanently enabled in cases where such detection fails or produces unintended behavior. Aside from RAWM, I'm only aware of Rapoo having this option on the VT3 Pro Max. Another interesting setting allows one to select different wireless modes, which are said to behave differently depending on the distance between mouse and dongle as well as further environmental factors (2.4 GHz band congestion, EMI, etc.). In my testing, I've found that this has a rather large effect on overall connection stability, as when using the "Mode 3" suggested by the software, I experienced no issues during several hours of usage, but on the next day, tracking frequently was suspended, which could only be fixed by switching to "Mode 1." Hence, while giving users a choice is good, I'd rather see this negotiated by the mouse automatically without any user input being necessary, as having to adjust this to avoid issues is more of a nuisance than a perk. Lastly, a setting called X-Polling is also present. This allows one to set polling rate (or, more specifically, report rate) to any value between 50 and 8000 in Hz. The issue with this is that conceiving of polling in terms of frequency is somewhat misleading, as USB specifications define intervals. Hence, when opting for a value such as "50 Hz," the mouse will simply alternate between intervals of 16 and 32 ms, resulting in severe temporal jitter. As such, I consider this option a gimmick at best, and would find it better if only intervals compliant with USB specifications were listed, such as 375 μs ("2666.67 Hz") or 1625 μs ("615.38 Hz")—i.e., multiples of frame size (125 μs). Somewhat ironically, such "native" steps cannot be set in the software as decimals are not supported. Another thing worth mentioning is that the software tracks every option set by the user to display a percentage indicating which portion of the userbase is using a specific setting. Unfortunately, there is no way to opt-out of this sort of telemetry, so one simply has to live with it.

In terms of performance, the ES21M does well. CPI deviation sits at +2.5%, which is easily corrected. A minor hurdle is posed by the software, as simply adjusting the slider won't work, given how the value will revert when switching to a different step. Hence, the step in question needs to be edited directly, and one also needs to be aware that non-native values will not be truncated to native ones. Native steps are multiples of 10, which is why the deviation present can be corrected fully. General tracking is fine on the ES21M, regardless of whether MotionSync is enabled or not, but this only applies under a specific condition. In both wired and wireless operation, the ES21M is limited to a 12-bit HID descriptor, which is why any report containing more than 2047 (+/-) counts will invariably result in a malfunction. In-game, this shows itself in the mouse simply not tracking. At lower CPI steps and higher polling rates, even fast swipes won't surpass this limit, but when using 1000 Hz in combination with something like 15,000 CPI, it becomes easy to trigger. Ultimately, most people likely won't even notice this, but on a technical level, there is simply no reason for this restriction to exist, and having a 30,000 CPI sensor when only a fraction of that is actually usable is somewhat pointless. Motion delay is on a competitive level on the ES21M. At 2000 and 4000 Hz in wireless operation, the ES21M is ahead of the Logitech G403 (control subject) by 0.7 ms without MotionSync, though enabling MotionSync will decrease this advantage. At 8000 Hz, the ES21M is only ahead by 0.5 ms, and it appears to be struggling in general at this polling rate, as tracking becomes rather messy. Additionally, there are occasional off-period polls, though aside from that, the target interval of 0.125 ms is averaged. Generally, polling is stable, instances of elevated noise notwithstanding. In terms of click latency, the ES21M delivers good but not outstanding numbers. The lowest average in wireless operation is achieved at 1000 Hz, where 1.7 ms can be measured when using the lowest debounce time value of 1 ms. Higher polling rates actually increase click latency, up to 2.0 ms at 8000 Hz. Higher debounce time values will scale linearly, and since slam-clicking, which describes inadvertent button actuation upon resetting the mouse, already occurs at 1 ms, opting for a higher value (and accordingly higher latency) may be necessary.

RAWM cites up to 85 hours of battery life at 1000 Hz and up to 21 hours at 8000 Hz. While the percentage-based battery life indicator found in the software isn't particularly accurate or reliable, it is sufficient for gauging expected battery life. Based on my testing, 20 hours at 8000 Hz indeed are plausible, which would translate to roughly 80 hours at 1000 Hz, give or take a few hours depending on sensor run mode. Given that a 300 mAh battery is used, this is in line with expectations and what the competition is able to muster. Charging isn't particularly speedy on the ES21M, and the charging cable makes matters worse, as it is both stiff and short. Hence, playing while charging is out of the question on the ES21M, and keeping the mouse charged at all times recommended.

Overall, the ES21M is a very solid contender not just for the SCYROX V8, but other budget releases as well. In terms of click latency, it cannot keep up with the best, but overall performance is still on a very good level. Of note in particular is the software, as it provides a wealth of options rarely found elsewhere, and the build quality displayed by the ES21M easily competes with releases in the $100 and more price bracket. Priced at $69.99 on Mechkeys.com, the ES21M provides excellent value for money, though given the choice, I'd probably opt for the white color variant, as it'll look less greasy than the black one. In any case, the ES21M earns our Budget award, and narrowly clutches our Recommended award as well.