Maxed-out HP z800 RAM: 384 GB of DDR3

In our lab we have 2 x HP z800 workstations. It is somehow ridiculous piece of hardware in terms of today standards. It’s got over 1 kW power supply, dual CPU motherboard and 12 memory slots. It is loud, draws loads of power and it’s got comparable gen 1 Intel CPUs as well as DDR3 – slow – memory. However it costs close to nothing and it is suitable for most small and mid applications.

Official documentation says that the maximum memory capacity is 192GB of DDR3, which is 12 x 16 GB sticks working at 800 MHz. If we are interested in higher speeds, then we go to 96GB at 1067 MHz using 16GB sticks or 1333 MHz at all 8GB sticks. Currently (2025) DDR5 memory works up to 8800 MT/s, so it is roughly 10 times faster, however with higher latencies. Much higher clock speeds compensates higher latencies.

Even though having 384GB of DDR3 is a lot for regular computing like application servers, mail servers, file storage, surveillance etc. At our lab we are using this server for data mining which includes Docker containers as well as database servers (various types).

However, we can put more memory than 192GB. We put as twice as much. Actually, installing Intel Xeon x5660 we would be able to handle as much as 288GB per CPU giving 576 GB of RAM in total. But we have only 12 memory slots and maximum capacity of memory module is 32GB. If we would have 18 memory slots then we would be able to put 576 GB, but with 12 memory slots we are able to place “only” 384GB.

HP z800 motherboard and Xeon CPUs take both unbuffered and registered DIMMs but not at the same time. So be sure to buy only one type, which I think it should be either 8500R or 10600R. Is is not worth to buy faster DIMMS as they still will be working at 800MT/s.

One important quirk regarding 32GB DDR3 memory sticks is that they are not recognized properly but dmidecode:

Same thing with lshw:

To access 12 x memory slots you need to remove double fan case. Moreover each memory module contains metal plate for heat dissipation. It can get really hot if put under heavy load.

BIOS says that everything is fine. All DIMMS are recongnized. Each Xeon x5660 and x5690 has 3 memory channels so be sure to place same type/speed/latency memory module in 2 consecutive memory slots. As per documentation there is preffered way to populate memory sticks in case of placing less than maximum number of them, so be sure to either place 12x same sticks or follow documentation to have the most close confiugration.

To try installing 384GB of RAM I was inspired by https://www.youtube.com/watch?v=AoS0kX82vs4 – it worked out! Finall note. Regular market price of 32GB DDR3 sticks varies between 50 and 300 PLN (10 – 70 EUR). However I was able to find 24 sticks (as I we have 2 such servers) for as low as 7 EUR per stick. Few years back it would be cost of 1600 EUR, today it was 160 EUR.

Thoughts on GPU thermal issues

I’ve been playing around with several devices in a context of running OpenCL code on them. They have one common thing which is excessive heat coming out of GPU and heatsink being unable to dissipate it. I start with MacBookPro3,1. It has NVIDIA 8600M GT, which is known to fail. I assume that it may be linked with overheating. Second example is design failure of Lenovo Thinkpad T420s which has built in NVIDIA NVS 4200M. This laptop has Optimus feature which in theory could detect if workload should be run on discrete or integrated GPU. Unfortunately enabling either Optimus or run-only-on discrete GPU causes extreme overheating up to 100 degress Celcius which makes this setup unusable (it slows down). Last example would be Lenovo Thinkpad T61 with NVS 140M. Contrary to previous examples, this one shows no issues with overheating itself, but is extremely fragile in terms of build quality. CPU has proper heatsink contact by means of 4 screws, but for unknown reason GPU which is 2 cm aside lacks of any screws and is dependent on separate metal clip which puts pressure on heatsink and has its own screw. I find it quite silly, because in case of thermal paste going bad or having loose this one screw it may completely damage GPU. Unscrewing just a little bit this one screw and temperature goes from 50 to 100 degrees risking fire I think….

So, back in a days when manufacturers tried to put dedicated GPU in compact laptops there were several examples of design flaws especially lacking proper heatsink and fans. Nowadays when discrete graphics are way more common on the market it is not uncommon to see several fans and huge blocks of metal giving away all this heat coming out from case because you run game or try to compute something in OpenCL.