WEBVTT 00:00:00.320 --> 00:00:07.120 Everyone knows that compared to a hard drive, an SSD is a great way to speed up 00:00:05.040 --> 00:00:14.480 your system. But what everyone doesn't know is that unless your SSD has a DRAM 00:00:11.360 --> 00:00:17.199 cache built into it, like this one for 00:00:14.480 --> 00:00:22.560 example, it can actually end up being slower than a mechanical hard disk. So 00:00:20.560 --> 00:00:29.279 that's why in the past we've always avoided those $20 SSDs 00:00:26.800 --> 00:00:37.360 until now. Meet the Fatty Dove Racing 120 gig SSD. 00:00:34.559 --> 00:00:42.660 We purchased this sucker for 20 bucks. So, who's the sucker? This sucker or 00:00:40.000 --> 00:00:45.770 this sucker? 00:00:48.800 --> 00:00:56.320 I know it shouldn't matter, but whenever I see really terrible branding, like to 00:00:54.000 --> 00:01:01.920 the point where you'd have had to be a child or an idiot to think that it makes 00:00:58.879 --> 00:01:04.320 any sense, it really does decrease my 00:01:01.920 --> 00:01:10.720 confidence in the finished product. You know, why is there an overweight bird on 00:01:07.920 --> 00:01:16.720 the box for this package? What is it about this that screams racing? I don't 00:01:14.320 --> 00:01:20.799 know. You don't know. Nobody knows. We're all just guessing. So, let's talk 00:01:19.119 --> 00:01:29.040 through the specs here. This is a 2 and a half inch drive with a an apparent 00:01:25.200 --> 00:01:31.840 128 mgby of DRAM cache. This is from the 00:01:29.040 --> 00:01:35.600 Amazon listing. Um, the verbiage though is lifted straight from the Wikipedia 00:01:33.680 --> 00:01:43.119 entry for other solid state drives. So, it says DRAM based and cash on buffer. 00:01:40.159 --> 00:01:50.960 But the thing is, the only other SSDs that we've seen at this price point are 00:01:46.000 --> 00:01:53.840 DRAMless, used, or renewed. And as far 00:01:50.960 --> 00:01:58.640 as I can tell, this thing is totally brand new in box. So, 00:01:57.280 --> 00:02:03.439 we also have this that came in the package. Oh, service policy. Thank you for 00:02:01.920 --> 00:02:07.920 purchase of our items. We're so glad that you selected us in our products. We 00:02:05.600 --> 00:02:13.360 offer a 5-year limited warranty. Love you, Dave. We got our SATA cable. That 00:02:10.479 --> 00:02:18.000 is actually a pretty good bonus. Most drives don't include a SATA cable 00:02:14.800 --> 00:02:19.599 anymore. So, we need our data and our 00:02:18.000 --> 00:02:24.879 power. And is this just going to boot right up for me, Anthony? Uh, if you press the power button. Yeah. 00:02:22.160 --> 00:02:28.160 Well, yeah. Minor details. Uh, you don't need to hit F8. 00:02:26.640 --> 00:02:32.239 Wait, is this just a sticker? Sure is. So, if the sticker is on it, you get 00:02:30.640 --> 00:02:36.720 plus 2 years of warranty. I don't know. Sudden power off recovery. That to me 00:02:35.040 --> 00:02:40.400 indicates that it would have a capacitor on it like a like an enterprise SSD. 00:02:39.120 --> 00:02:46.080 That's what it sounds like. It can't possibly have that. Also, I 00:02:43.920 --> 00:02:50.560 love this sentence right here. Thank you for selecting us. Good luck. So, first 00:02:48.720 --> 00:02:54.879 test is what? Just initializing it. Yeah, we need to initialize it. 00:02:51.920 --> 00:02:58.480 New simple volume. Well, the capacity is as advertised. Now, there's some more 00:02:56.640 --> 00:03:03.599 problems here. This says serial ATA revision 2.6, six, but it also says it's 00:03:01.519 --> 00:03:05.599 SATA 6 Gbit per second. And it said that on the drive as well. 00:03:04.959 --> 00:03:14.400 Yeah. So, apparently it doesn't even support trim and is SATA 2 according well 00:03:12.640 --> 00:03:18.000 depending on which spec you believe. Well, we'll know soon enough because 00:03:15.840 --> 00:03:21.040 Crystal Disc Mark will tell us exactly what kind of sequential speeds we can 00:03:19.519 --> 00:03:25.360 expect from this thing. You know, in some ways this like rando cheap hardware 00:03:23.680 --> 00:03:29.040 is far more interesting than the stuff from manufacturers that are reputable 00:03:27.360 --> 00:03:33.360 enough that you can trust what they say. Yeah, you never know what you're going to get. 00:03:30.480 --> 00:03:36.879 Look at that. So, it's SATA 3 6 GB per second even though there's information 00:03:35.120 --> 00:03:41.280 right on the box that would indicate that it's not 00:03:39.280 --> 00:03:45.599 what they put a dove on their packaging. So, how how can we rely on them to know 00:03:44.080 --> 00:03:51.840 exactly what's in the box if they clearly had no idea what they were doing on the outside of it? Out of the gate, 00:03:49.280 --> 00:03:57.040 I'm actually kind of impressed. I mean, saturating the SATA 3 6 gig per second 00:03:54.159 --> 00:04:02.959 interface has not been a feat for years and years at this point. But the thing 00:03:59.040 --> 00:04:05.200 is, this is only a 120 gig SSD. So, 00:04:02.959 --> 00:04:10.319 what's weird about that is that you know how higher capacities of a given SSD 00:04:07.840 --> 00:04:14.319 tend to be rated for more performance. The reason for that is that to get those 00:04:12.480 --> 00:04:18.720 higher capacities, in some cases, they'll put more nan flash chips onto 00:04:17.040 --> 00:04:22.639 the PCB, which means that you get to take advantage of more parallelization 00:04:20.799 --> 00:04:27.840 as you're writing to the flash and reading from it, boosting your 00:04:24.000 --> 00:04:32.000 performance. Well, it's been years since 00:04:27.840 --> 00:04:34.160 120 gig SSDs have made any sense to make 00:04:32.000 --> 00:04:40.400 because the capacities of the individual nan flash chips have become so high that 00:04:37.520 --> 00:04:44.320 in order to even build a 120 gig SSD, you'd only need like one chip on there 00:04:42.560 --> 00:04:52.160 and the performance would absolutely suck unless it's fine. 00:04:48.960 --> 00:04:53.440 Okay, so the enigma deepens then. What 00:04:52.160 --> 00:05:01.199 did you have in mind for us next, Anthony? Uh, well, we could do the file copy 00:04:56.880 --> 00:05:03.040 test, but um I did create some tests 00:05:01.199 --> 00:05:06.800 specifically for this in Linux. Okay, so I should restart. Yep. 00:05:05.120 --> 00:05:15.039 All right. So, okay, hit me. Pop open two terminals and we'll run 00:05:10.560 --> 00:05:17.199 some uh scripts that I made. Sudo slash 00:05:15.039 --> 00:05:18.960 disc test. All right. So, uh that one then 00:05:18.240 --> 00:05:23.280 that is the one 111 gig SSD. Yeah, it does a security 00:05:21.360 --> 00:05:27.280 erase first and now it's doing continuous looping of 1 GB right and 00:05:25.919 --> 00:05:31.440 then 1 GB read. Okay. And then it tells us what the latency is 00:05:29.600 --> 00:05:34.320 after a thousand writes of average. Oh, that took a lot longer already. Oh, 00:05:33.280 --> 00:05:39.360 it's going up. Oh boy. We went from 2.3 seconds to 7 seconds 00:05:37.440 --> 00:05:42.720 already. The reads are still pretty consistent, but the rights have 00:05:41.520 --> 00:05:51.440 skyrocketed. This is TLC. Okay, that's fair. All right. 00:05:46.479 --> 00:05:55.039 So, the other one, um, do sudo file copy 00:05:51.440 --> 00:05:57.120 off. What? I swear to God, the way that 00:05:55.039 --> 00:06:05.360 WhatsApp sends you constant notifications that you're logged out 00:06:00.400 --> 00:06:08.080 P space mount and hit enter. That all 00:06:05.360 --> 00:06:12.160 looks correct. And that'll show us then over here what our sequential read and 00:06:10.400 --> 00:06:16.319 writes look like and as well as especially the right latency. 00:06:14.000 --> 00:06:20.560 Got it. So that'll tell us how effective that DRAM cache is, 00:06:18.240 --> 00:06:24.479 as well as how effective the NAND is at being erased. 00:06:21.680 --> 00:06:27.919 This is interesting. So the read test is still taking about the same amount of 00:06:26.160 --> 00:06:34.720 time it was at the start, but we're getting spikes in the right test where 00:06:30.400 --> 00:06:36.560 it's jumping up to over 15 seconds. We 00:06:34.720 --> 00:06:41.039 just finished our first run of the sequential file copy test and it's 00:06:38.479 --> 00:06:46.240 getting warm now. And this is interesting. The read test is no longer 00:06:44.080 --> 00:06:51.520 finishing consistently in just two seconds. It's actually drawn almost 00:06:49.039 --> 00:06:53.919 level with the right test which has gone down. 00:06:52.160 --> 00:07:00.960 The right latency though has gone up pretty significantly. Quintupal or as much as 8x. It's all 00:06:57.680 --> 00:07:02.800 over the place. Okay. Does it have a DRM 00:07:00.960 --> 00:07:06.800 cap? No, don't tell me. We'll find out after. So now what? 00:07:05.120 --> 00:07:11.520 Well, now we can graph those results, but we could also throw on one of our 00:07:08.800 --> 00:07:16.240 DRMless SSDs. Oh, let's see. See if that uh that makes a difference. 00:07:13.599 --> 00:07:17.440 Okay. Uh so reboot. Do you want to look at that output 00:07:17.120 --> 00:07:23.199 first? Yeah. You want to look at the output first? Ah ltar.com. Oh, thank you. Did you 00:07:21.599 --> 00:07:28.319 really build that into this benchmark? I built many things into the benchmark 00:07:24.720 --> 00:07:30.880 that you have not seen yet. 00:07:28.319 --> 00:07:35.360 That moment when your program is like 90% Easter eggs, 10% copying files back 00:07:34.080 --> 00:07:38.479 and forth. Dear lord. Oh my god, Anthony. 00:07:38.160 --> 00:07:43.759 What? How do I shut it down? Same way you just go to go to leave 00:07:41.680 --> 00:07:48.160 leave. Why is it called leave? I mean that's almost as stupid as Windows being 00:07:45.840 --> 00:07:50.560 start to turn it off. So just the same commands again. 00:07:49.599 --> 00:07:56.479 Same commands again. So we'll start with the A data drive. That is 00:07:54.240 --> 00:08:00.240 I want to say similar and then I can start this one. Yeah. 00:07:58.000 --> 00:08:02.160 Okay. Let's see what happens. Now this product page is just going to be lies, 00:08:01.840 --> 00:08:07.599 right? Do you want to look at it? I mean we've got some time. I'm pretty sure most of 00:08:05.759 --> 00:08:12.319 this is lifted either from Wikipedia or like reviews of other unrelated SSDs. 00:08:09.919 --> 00:08:16.240 Single level cell cache technology. I don't think so. 00:08:13.280 --> 00:08:20.800 Uh definitely not. SLC hasn't been around even in the enterprise for many 00:08:18.720 --> 00:08:25.199 years. There's even conflicting information here. It takes 10 seconds 00:08:23.039 --> 00:08:30.319 from the start to the desktop. Fatty Dub 240 gig solidstate drive starts up in 00:08:27.039 --> 00:08:33.440 about 15 seconds. They're this far away 00:08:30.319 --> 00:08:34.240 from each other. Optical storage read 00:08:33.440 --> 00:08:38.240 speed. What? What are you even talking about? Do we 00:08:36.479 --> 00:08:44.320 have a result here? No, not just yet. But I not quite yet, but I think sure this is 00:08:41.200 --> 00:08:47.920 a DRAMless SSD. Wait, 00:08:44.320 --> 00:08:51.040 it actually might not be. This a data is 00:08:47.920 --> 00:08:54.000 doing even worse. Wait, no. Better in 00:08:51.040 --> 00:08:58.720 latency for the most part, but the actual test completion times are much 00:08:56.080 --> 00:09:04.519 worse. That actually would make some sense if the Fatty Dove only has a 00:09:00.800 --> 00:09:04.519 couple of chips in it. 00:09:05.519 --> 00:09:11.440 Thankfully, Fatty Dove is more resistant 00:09:08.959 --> 00:09:15.600 to shock and vibrations. Can fly, but like not quite fly. It just sort of like 00:09:14.320 --> 00:09:19.839 falls slowly, you know? I think it's just like a chicken. Yeah. Yeah, chickens can do that. Have 00:09:18.560 --> 00:09:22.640 you ever thrown a chicken? No. Fun fact, I actually had a pet 00:09:21.760 --> 00:09:28.320 chicken when I was a kid. Me, too. The chicken was the one that all the other chickens picked on. Yeah. So, we 00:09:26.240 --> 00:09:33.600 took the chicken away and we raised it and it became nice and big. And then one 00:09:30.720 --> 00:09:36.480 day we had to give it back for reasons. Did your parents make you eat it though? 00:09:35.600 --> 00:09:45.200 No, it was cuz that's what happened with my pet goat. Oh, run three finished. Wow. 00:09:41.839 --> 00:09:48.080 250 seconds by the last run here. This 00:09:45.200 --> 00:09:50.959 one got absolutely creamed. All right, then. 00:09:48.480 --> 00:09:55.200 Okay, point to Fatty Dub, I guess. Control. Well, I mean, 00:09:52.640 --> 00:10:00.560 for throughput. Yeah, beating the worst is not exactly an achievement. So, round 00:09:58.240 --> 00:10:03.519 two. Uh, this one right here. Should I just start ripping this thing open? 00:10:02.080 --> 00:10:10.959 Do you want to avoid that warranty? Oh, yeah. Open. Wow, that sticker. Stop. 00:10:08.399 --> 00:10:13.680 You're done. Oh, and it's bent. 00:10:12.320 --> 00:10:21.600 Yep. You bent your fatty dove. Yeah. Wow. That is That is not a lot of 00:10:19.120 --> 00:10:24.320 PCB. Tell me something, Anthony. Are you finding this out for the first time or 00:10:22.959 --> 00:10:28.880 did you have one of these open already? I did not have one of these open already. I should have gotten one and 00:10:27.120 --> 00:10:32.800 opened one, but this is actually very fascinating to me right now. 00:10:30.399 --> 00:10:36.079 Now, I have seen SSDs that have taken a similar approach with a full size 2 and 00:10:34.720 --> 00:10:40.160 1/2 in enclosure. And I've actually seen ones that are like as small as just 00:10:38.160 --> 00:10:45.920 right here. Okay. So, based on looking at the back 00:10:43.200 --> 00:10:50.880 of the PCB, I'd say that's our controller right there. And I'm guessing 00:10:48.480 --> 00:10:57.120 no DRAM cache cuz it looks like there's just one chip right there. 00:10:53.839 --> 00:11:03.040 It does have a DRAM cache. It's right 00:10:57.120 --> 00:11:05.920 here. No way. So that seems to be a 00:11:03.040 --> 00:11:10.079 Silicon Motion controller. We're going to have to check this Samsung part 00:11:07.440 --> 00:11:18.160 number to find out if the 128meg cache is correct. And then as we discussed, 00:11:14.320 --> 00:11:21.920 there is in fact only a single nan flash 00:11:18.160 --> 00:11:24.480 chip on this PCB. But based on how far 00:11:21.920 --> 00:11:29.519 flash has come in the last few years, I guess you can get adequate enough 00:11:26.240 --> 00:11:30.959 performance for a SATA SSD out of just a 00:11:29.519 --> 00:11:36.640 single chip, which I guess kind of makes sense cuz you know a max capacity NVMe 00:11:34.320 --> 00:11:41.279 drive is only going to have four chips on it with a lot of the high performance 00:11:38.079 --> 00:11:44.560 ones having just two. So SATA is like a 00:11:41.279 --> 00:11:46.720 fraction of that speed. Wow, I'm 00:11:44.560 --> 00:11:53.440 impressed. Okay, let's find out how much cache it actually has, though. Oh, this 00:11:48.320 --> 00:11:55.839 benchmark is done and the Micron SSD is 00:11:53.440 --> 00:11:58.959 definitely better, but not necessarily because it has a DRM cache. Probably 00:11:57.680 --> 00:12:04.320 just cuz it's got better engineered firmware and a better controller. Okay, I want to I want to check these part 00:12:01.920 --> 00:12:11.120 numbers now. I think that's an E D RAM chip. DDR3 SD RAM 1 GB. Okay, so 128 00:12:08.639 --> 00:12:15.440 megs. Surprisingly then, as broken as the packaging and product page were, the 00:12:13.440 --> 00:12:20.240 main controversial element of this thing, whether or not it had a DRAM 00:12:17.200 --> 00:12:22.000 cache, ended up being true. Uh although 00:12:20.240 --> 00:12:26.880 with that said, the power loss protection uh did not end up being 00:12:24.079 --> 00:12:32.720 present. There's no bank of capacitors here that looks anything like that. So 00:12:29.279 --> 00:12:34.560 that that is clearly not accurate. Why 00:12:32.720 --> 00:12:38.000 why would they say that? Is it to do with the fact that compared to a hard 00:12:36.560 --> 00:12:44.399 drive, you don't have to worry about if you lose power, the head parking? 00:12:41.279 --> 00:12:47.519 Oh, or it could be that they mean that 00:12:44.399 --> 00:12:49.920 compared to something like just a D a 00:12:47.519 --> 00:12:51.920 RAM drive, the data is held persistently, 00:12:51.120 --> 00:12:56.240 right? Right. Maybe it's it's like volatile. 00:12:54.160 --> 00:13:00.639 Yeah. Non nonvolatile storage. I don't know. I don't know what they were going 00:12:57.200 --> 00:13:02.720 after there. But for 20 bucks, actually 00:13:00.639 --> 00:13:05.920 not bad. for longevity, right? But 00:13:03.920 --> 00:13:09.360 yeah, can't can't speak to that. But hey, 5year warranty, right? So, thanks 00:13:07.760 --> 00:13:14.720 for watching, guys. If you enjoyed this little adventure in exploring weird 00:13:12.480 --> 00:13:19.200 cheap hardware off of AliExpress, then maybe check out uh our oh our show and 00:13:17.760 --> 00:13:23.200 tell where everyone bought the weirdest hardware they could possibly find and 00:13:20.800 --> 00:13:27.839 then see who had seed saw, who had whatever. It was weird. Go check it out. 00:13:25.040 --> 00:13:27.839 Thanks Bluff.