WEBVTT

00:00:00.160 --> 00:00:06.560
apple's mac studio set a high bar for performance in a compact workstation but

00:00:04.720 --> 00:00:12.160
hp is here to say no way san jose or cupertino neat

00:00:12.719 --> 00:00:20.320
what is it glued it the z2 mini g9

00:00:16.560 --> 00:00:23.359
according to hp it is the world's most

00:00:20.320 --> 00:00:26.240
powerful mini workstation but then

00:00:23.359 --> 00:00:30.480
also according to hp that's a big load of horse poo

00:00:27.439 --> 00:00:33.040
so what exactly is it then

00:00:30.480 --> 00:00:40.000
well it's a sleek mini workstation boasting up to a core i9 12900k

00:00:36.239 --> 00:00:42.559
and NVIDIA a2000 graphics and those are

00:00:40.000 --> 00:00:47.840
some very impressive specs i mean clearly hp is bringing the heat

00:00:44.879 --> 00:00:51.680
the question is then can they keep cool under the pressure

00:00:49.520 --> 00:00:56.000
like i keep cool when segwaying to our sponsor

00:00:52.879 --> 00:00:58.399
jackery their explorer 2000 pro portable

00:00:56.000 --> 00:01:03.120
solar generator provides a massive 2000 watt hour capacity charges up to eight

00:01:00.640 --> 00:01:07.280
devices at once and more on prime day you can save 900

00:01:05.199 --> 00:01:11.280
on an explorer 2000 bundle learn more below or at the end of the video hp has

00:01:09.760 --> 00:01:15.680
gotten a pretty rough ride on this channel and deservedly so nearly every

00:01:14.080 --> 00:01:20.799
fascinating innovation they've brought to us has been spoiled by poor pricing

00:01:18.720 --> 00:01:25.040
arbitrary hardware limitations and anti-diluvian build quality sometimes

00:01:23.119 --> 00:01:31.040
all three they just can't seem to point the gun away from their foot maybe this

00:01:29.040 --> 00:01:35.600
time will be different but maybe it won't looking at the z2 mini g9's i o

00:01:33.520 --> 00:01:41.520
we're off to a reasonably good start here they offer a range of display usb

00:01:38.720 --> 00:01:49.360
Thunderbolt and network connectivity and they provide both a PCIe 4.0 by 8 half

00:01:45.439 --> 00:01:51.360
height slot with GPU support and a flex

00:01:49.360 --> 00:01:57.759
i o slot that can accommodate anything from extra display outputs or usb ports

00:01:54.880 --> 00:02:03.360
to serial ports though a likely use for it will be to remedy the inexcusable in

00:02:00.640 --> 00:02:08.560
my opinion absence of multi-gigabit networking in the base spec hp this is a

00:02:06.399 --> 00:02:13.040
workstation product and even at my purchase volumes two and a half gig

00:02:10.319 --> 00:02:17.280
Ethernet costs four dollars per chip the case can lie flat or be vasa mounted and

00:02:15.680 --> 00:02:21.440
if you want to get really fancy you can pick up a vertical mount and flip the

00:02:19.440 --> 00:02:25.760
logo around wow the front is all mesh for better airflow

00:02:23.599 --> 00:02:28.879
and hp helpfully points out that if you really wanted you could buy a whole

00:02:27.360 --> 00:02:33.760
gaggle of them and put them on a rack mount shelf neat first though we need to

00:02:31.520 --> 00:02:37.840
drag race it against the other most powerful mini workstation and see

00:02:36.160 --> 00:02:43.200
if there's any reason to buy one of these things let alone seven of them as

00:02:40.720 --> 00:02:48.640
configured rz2 mini g9 is nearly identically priced to the m1 ultramax

00:02:45.519 --> 00:02:51.680
studio spec for spec

00:02:48.640 --> 00:02:55.120
let's begin in apple's weakest category

00:02:51.680 --> 00:02:56.080
gaming hp pulls out an easy way wait

00:02:55.120 --> 00:03:01.840
what in shadow of the tomb raider the m1

00:02:58.400 --> 00:03:05.440
ultra beats it with the cheaper max tier

00:03:01.840 --> 00:03:08.800
tying the hp workstation boy this is not

00:03:05.440 --> 00:03:11.680
a good start and this trend continues in

00:03:08.800 --> 00:03:16.800
total war warhammer where again hp is bested by the mighty mac ultra the story

00:03:14.560 --> 00:03:21.440
flips with counter strike but this is mostly due to cs go running on opengl on

00:03:20.080 --> 00:03:25.599
macOS now obviously game compatibility isn't

00:03:23.920 --> 00:03:29.760
going to favor the mac but given that we're evaluating these machines as

00:03:27.440 --> 00:03:34.239
workstations our true objective in running games is to find the performance

00:03:31.760 --> 00:03:40.000
potential of the hardware and it is clear that hp's tiny workstation is

00:03:37.200 --> 00:03:45.680
limited by NVIDIA's a2000 a cute little GPU that can be thought of as a 3050 but

00:03:43.280 --> 00:03:50.239
on a 70 watt power diet with a whopping 12 gigs of vram and certified drivers

00:03:48.159 --> 00:03:56.000
for professional applications for high resolution video editing and 3d modeling

00:03:52.959 --> 00:03:58.080
a gaming GPU it clearly ain't so let's

00:03:56.000 --> 00:04:05.040
move on to productivity where surely the story will stay exactly the same in

00:04:01.840 --> 00:04:08.400
puget bench premiere hp loses to apple

00:04:05.040 --> 00:04:09.760
again though in photoshop hp does score

00:04:08.400 --> 00:04:13.680
its second win though it should be noted that puget

00:04:11.439 --> 00:04:19.359
bench relies on rosetta for photoshop on macOS so it's not really apples to

00:04:16.880 --> 00:04:23.680
apples after effects is another win for apple given we couldn't get the

00:04:20.959 --> 00:04:28.880
benchmark to run on rz2 mini g9 so we can just skip that i guess

00:04:25.840 --> 00:04:31.360
and as for blender well in CPU rendering

00:04:28.880 --> 00:04:37.440
it loses against the ultra in our bmw test and in gooseberry which raises some

00:04:34.800 --> 00:04:43.680
very interesting questions because we had a similarly configured test bench

00:04:40.000 --> 00:04:47.120
in our mac studio review that did win

00:04:43.680 --> 00:04:47.120
with a core i9

00:04:47.199 --> 00:04:55.360
in our GPU render testing rendering with optics wins every time but with cuda the

00:04:52.400 --> 00:05:00.240
mac ultra pulls ahead again as for synthetic productivity benchmarks

00:04:57.840 --> 00:05:04.000
in cinebench hp wins in single threaded performance but then fumbles in

00:05:02.000 --> 00:05:08.560
multi-threaded performance and the same goes for geekbench signaling the

00:05:05.680 --> 00:05:12.560
existence of a clear trend here core i9 12900k

00:05:11.120 --> 00:05:18.400
plus heat name a more iconic duo i'll wait oh yeah

00:05:16.720 --> 00:05:23.840
i guess the wan hoodie and when sweatpants from lttstore.com

00:05:21.759 --> 00:05:30.160
now it should be noted that our m1 ultramax studio is decked out with

00:05:26.440 --> 00:05:31.919
128 gigs of RAM but that's unlikely to

00:05:30.160 --> 00:05:35.840
heavily factor into the tests that we ran today

00:05:33.199 --> 00:05:40.960
what that means is that hp must have screwed something up in here

00:05:38.479 --> 00:05:45.680
to find out what we have to take a bit of a deeper dive inside actually

00:05:43.840 --> 00:05:50.080
we won't even really need to do that because like you we did a double take

00:05:48.560 --> 00:05:54.240
when we saw they were trying to pack a core i9 into this tiny three and a half

00:05:52.560 --> 00:06:01.680
liter enclosure i mean the mighty nhd15 can barely handle this

00:05:57.360 --> 00:06:05.520
CPU and it alone is four liters in

00:06:01.680 --> 00:06:07.520
volume did hp engineer a magic 240 watt

00:06:05.520 --> 00:06:11.600
cooling solution out of unicorn horns and leprechaun blood

00:06:09.120 --> 00:06:15.479
nope they didn't even try in fact you can see it right here there's a sticker

00:06:13.280 --> 00:06:22.000
on the cooler that says 125 watts so naturally the poor thing

00:06:19.280 --> 00:06:27.680
throttled hard throughout our testing in cinebench the z2 performs a whopping 30

00:06:25.039 --> 00:06:32.880
percent worse than the exact same chip on our test bench in blender our render

00:06:30.240 --> 00:06:38.000
times increased between 20 and 50 percent over our test bench and premiere

00:06:35.280 --> 00:06:42.800
pro we see 10 percent lower scores in puget bench with similar losses in

00:06:40.319 --> 00:06:47.759
geekbench this performance hit even bleeds into gaming where we see 10

00:06:45.280 --> 00:06:53.759
losses for the hp workstation in tomb raider warhammer and cs go

00:06:51.280 --> 00:06:57.520
we also took forza horizon 5 for a spin and saw

00:06:54.800 --> 00:07:01.840
again a 10 hit and even more damningly running the

00:06:59.919 --> 00:07:07.280
benchmark for an extended period of time drops the z2 mini g9's result even

00:07:04.560 --> 00:07:11.840
further why would that be you might ask surely the CPU is already thermally

00:07:09.520 --> 00:07:16.479
throttled and you're right but as it turns out thanks to hp's

00:07:14.560 --> 00:07:22.880
brilliant case design which provides next to no clearance for the GPU cooler

00:07:19.199 --> 00:07:26.160
the GPU overheats too that's right even

00:07:22.880 --> 00:07:28.560
our tiny little 70 watt GPU is

00:07:26.160 --> 00:07:34.479
throttling to the point where longer GPU renders complete 25

00:07:31.280 --> 00:07:36.560
faster on our bench compared to the hp

00:07:34.479 --> 00:07:42.000
it's just sad even cranking the GPU fan to a hundred

00:07:39.520 --> 00:07:47.280
percent which is well beyond the factory curve only delays the inevitable

00:07:44.319 --> 00:07:50.639
throttling and as for the CPU well we tried some simple adjustments to their

00:07:49.039 --> 00:07:55.360
cooling solution in an attempt to prevent hot air from recirculating

00:07:52.800 --> 00:07:59.520
inside the case and we were able to marginally improve our scores but what

00:07:57.440 --> 00:08:04.000
we really need is a way to adjust the fan curve something that hp

00:08:02.080 --> 00:08:10.560
either through their performance advisor software or the BIOS shockingly does not

00:08:07.840 --> 00:08:15.599
allow and it needs to be adjusted look at this we started up a benchmark this

00:08:12.879 --> 00:08:18.960
thing is running full bore right now and

00:08:17.039 --> 00:08:26.000
it's dead silent the fans didn't even spin up until after

00:08:22.560 --> 00:08:28.479
the CPU had already thermally throttled

00:08:26.000 --> 00:08:34.640
we tried every performance mode from quiet to high performance and they all

00:08:31.360 --> 00:08:37.039
behaved basically the same the only

00:08:34.640 --> 00:08:42.240
possible solution was to set the idle fan speed to 100 percent which

00:08:39.360 --> 00:08:49.040
unfortunately locks it at maximum which unfortunately is 7 000 RPM on this model

00:08:46.320 --> 00:08:52.720
which is loud as all hell and

00:08:49.920 --> 00:08:54.880
adding insult to injury the machine still

00:08:53.760 --> 00:08:59.760
throttles to demonstrate that this is a cooling

00:08:57.360 --> 00:09:09.200
problem rather than a spec problem we actually swapped out our core i9 with a

00:09:02.959 --> 00:09:13.040
core i7 12700k and in this exact system

00:09:09.200 --> 00:09:14.880
that i7 actually beat the i9 in

00:09:13.040 --> 00:09:20.080
rendering premiere and in cs go

00:09:16.880 --> 00:09:23.440
even our less power hungry CPU though it

00:09:20.080 --> 00:09:26.800
still throttled partially because hp's

00:09:23.440 --> 00:09:29.120
cut tang power limiting made yet another

00:09:26.800 --> 00:09:35.200
surprise appearance so sorry i gotta turn this off hp not only equips their

00:09:32.320 --> 00:09:40.959
workstation with insufficient cooling their workstation class motherboard

00:09:37.519 --> 00:09:43.680
can't even power the hardware anyway

00:09:40.959 --> 00:09:48.480
i mean knowing that hp employs some incredibly talented engineers it's hard

00:09:46.320 --> 00:09:52.000
for me to imagine what goes on behind the scenes for a train wreck like this

00:09:50.399 --> 00:09:56.320
to make it to market unlike LMG where you can watch the train

00:09:54.160 --> 00:10:00.240
wreck in slow motion behind the scenes on Floatplane.com rather than just

00:09:58.399 --> 00:10:05.120
imagining it let's come back around to hp's claim though that the z2 mini g9 is

00:10:02.880 --> 00:10:10.000
the fastest mini workstation in the world we called that horse plop before

00:10:07.920 --> 00:10:15.279
we even knew that the machine throttled like an aging white dude on a harley why

00:10:12.800 --> 00:10:20.720
because their qualification for that claim is that it is

00:10:17.200 --> 00:10:23.040
based on hp's internal analysis of

00:10:20.720 --> 00:10:28.079
non-gaming small form factor desktop workstations with between three to 20

00:10:25.600 --> 00:10:32.800
leaders a minimum three isb certifications configurable professional

00:10:30.320 --> 00:10:39.600
graphics and a dedicated workstation brand as of march 2022 good gravy i mean

00:10:36.959 --> 00:10:44.560
by that kind of logic based on my analysis i have the most subscribers on

00:10:42.320 --> 00:10:49.279
youtube of people with at least eight channels that make tech videos based in

00:10:46.720 --> 00:10:54.560
surrey canada and we haven't even started on the price much like the mac

00:10:51.839 --> 00:10:59.279
studio this thing is very expensive i mean i get it it's a business system so

00:10:56.959 --> 00:11:03.440
there is a premium and you do get some benefits from it the parts have

00:11:01.120 --> 00:11:08.959
undergone extra reliability testing it uses a workstation chipset ecc memory

00:11:06.560 --> 00:11:16.720
isv certification you know the whole corporate kit in caboodle but charging

00:11:12.279 --> 00:11:18.959
1100. for a 12 900k that is 500 over

00:11:16.720 --> 00:11:22.800
retail for the privilege of running it throttled that's just insulting

00:11:21.760 --> 00:11:28.240
hp this could have been a very sleek

00:11:25.360 --> 00:11:33.200
capable workstation and for less powerful components it still could be an

00:11:30.640 --> 00:11:39.040
excellent fit for your business but as configured it simply shouldn't exist

00:11:37.360 --> 00:11:43.040
you know what should though our sponsor jackery whether you're

00:11:41.360 --> 00:11:46.399
looking to make margaritas on the beach living that van life or preparing to

00:11:44.720 --> 00:11:50.240
leave civilization the checkered explorer 2000 pro portable solar

00:11:48.320 --> 00:11:55.839
generator has all the juice you need to keep all of your devices energized the

00:11:52.320 --> 00:11:57.920
massive 2000 watt hour capacity and 2200

00:11:55.839 --> 00:12:02.560
watt output allows for eight devices to be plugged in simultaneously with a 100

00:12:00.560 --> 00:12:07.120
charge from an ac wall outlet in just two hours you can get in and out before

00:12:05.120 --> 00:12:11.440
anyone has even noticed you're there or pair your power station with an array of

00:12:08.800 --> 00:12:15.200
jackery solar saga 200 watt solar panels for a full charge in as little as two

00:12:13.279 --> 00:12:19.040
and a half hours and rid yourself of the need to even step a foot indoors again

00:12:17.440 --> 00:12:24.560
checkery has a special prime day offer coming up on july 12th where you can

00:12:20.959 --> 00:12:27.360
save 900 on the explorer 2000 pro and

00:12:24.560 --> 00:12:31.040
two solar sega 2000 watt solar panels check out the jackery explorer 2000 pro

00:12:29.360 --> 00:12:34.560
at the link down below and make sure you bookmark their amazon page so you're

00:12:32.800 --> 00:12:38.160
ready for that sweet prime day offer thanks for watching hp seriously you

00:12:36.560 --> 00:12:44.959
guys need to get your stuff together everyone else if you enjoyed this bloodshed why not check out our video on

00:12:41.279 --> 00:12:48.160
hp's omen 45l pre-built it

00:12:44.959 --> 00:12:48.160
also had problems