32GB RAM being maxed out and page file going crazy on SSD; Get a dedicated HDD?


Recommended Posts

SSD is the perfect place for a pagefile.

 

If anything, get more ram and get another SSD, dedicated for the pagefile.

 

 

 

I've never used the program in question, but check if it's got options for working drives or scratch disks.  In Photoshop, you can specify the drive/folder for it to use as a 'paging' area

Hello,

SSD is the perfect place for a pagefile.

 

If anything, get more ram and get another SSD, dedicated for the pagefile.

 

 

 

I've never used the program in question, but check if it's got options for working drives or scratch disks.  In Photoshop, you can specify the drive/folder for it to use as a 'paging' area

The opinions back and forth are amazing.

I should have made a poll.

Get a dedicated HDD? for page file?

got a dedicated RAM-disk, a ram disk that pretend to be a HDD at hardware level,

not a software based one that consuming your main RAM.

 

no more spinning-noise of traditional HDD or any possible hardware failure due to physical shock like an earthquake,

no more worry about NAND cells life cycles or financial burden replacing your SSD at regular intervals,

no other hassle configuring the OS, as OS would just assumes that just another ready-to-use storage devices.

but its a niche product i only found the small capacity one (only 64GB, and thats using ancient DDR2),

haven't seen a big capacity (min 512 GB), commercial RAM-drive anywhere, yet.

 

Hello,

Now this is a way better post! :)

I personally just do IT; I dont work with Solidworks and didnt even see it till about 6 months ago. When building this machine which was geared towards it, I read about softtweaks but since I didnt understand what my collegues need or want (and of course, I dont know or care what the program does), I went a hardware route. Maybe this would be a good time for software tweaks like you mentioned.

Well, since most of you are bent on limiting the program's RAM, I guess when we do this (upgrade the RAM) Ill go ahead and limit it. I gotta read on how to optimize it for Simulation AND also if it affects my collegues' needs.

This is the core of your problem... If you have no care to know how the program you're building the hardware for utilizes the hardware then you're going to end up wasting resources and not getting the results expected...

 

You need to understand the program your optimizing the hardware for to get them to perform optimally together... There is a reason large SQL DBs are run from 15K RPM SAS drives and not consumer grade 7.2K RPM SATA drives, for instance. Knowing the limits of the hardware along side the software demands for your usage case is important. You don't need SAS drives to push SQL server on your local machine with just you, but the same isn't true for a 10K concurrent user system powering a corporate DB holding TBs of data.

Hello,

This is the core of your problem... If you have no care to know how the program you're building the hardware for utilizes the hardware then you're going to end up wasting resources and not getting the results expected...

 

You need to understand the program your optimizing the hardware for to get them to perform optimally together... There is a reason large SQL DBs are run from 15K RPM SAS drives and not consumer grade 7.2K RPM SATA drives, for instance. Knowing the limits of the hardware along side the software demands for your usage case is important. You don't need SAS drives to push SQL server on your local machine with just you, but the same isn't true for a 10K concurrent user system powering a corporate DB holding TBs of data.

The program I built the hardware for only asked initially for CPU...because I looked for best optimization.

As a matter of fact, the RAM inside that PC is just some OEM POS. We used it because it was already here and because RAM was not a issue at the time.

That being said, I have to look up some tweaks for the program before hand of buying a storage device of some kind and since this thread is shifting left and right.

I've never used SolidWorks myself, but the Mechis here use it. I looked it up briefly and it appears to me that the excessive ram usage is the result of the number of nodes in the mesh in conjunction with the simulations done on that mesh (flow or otherwise). Seems to me you aren't going to be able to limit the memory without reducing the the mesh nodes or changing how the simulation is done. I don't personally know if the latter is even possible as I'm not a domain expert, but the former should be. I'm not really sure if this is a solution though because perhaps the complexity of the mesh is needed. In that case you are probably out of luck without an additional disk or ssd for scratch.

 

Though many people here are saying it is a software issue, they aren't taking into account that this is simulation software and the memory footprints of such things can be absolutely huge. We run hardware simulators here that eat 128+GB of physical memory, can run for days or weeks, and can fill up an entire disk with logs in just one run. It seems to me that depending on what exactly you are simulating, SolidWorks could be thrown on a server with 256GB of physical memory, TB of swap, and then set to run for weeks or a month.

Hello,

I've never used SolidWorks myself, but the Mechis here use it. I looked it up briefly and it appears to me that the excessive ram usage is the result of the number of nodes in the mesh in conjunction with the simulations done on that mesh (flow or otherwise). Seems to me you aren't going to be able to limit the memory without reducing the the mesh nodes or changing how the simulation is done. I don't personally know if the latter is even possible as I'm not a domain expert, but the former should be. I'm not really sure if this is a solution though because perhaps the complexity of the mesh is needed. In that case you are probably out of luck without an additional disk or ssd for scratch.

 

Though many people here are saying it is a software issue, they aren't taking into account that this is simulation software and the memory footprints of such things can be absolutely huge. We run hardware simulators here that eat 128+GB of physical memory, can run for days or weeks, and can fill up an entire disk with logs in just one run. It seems to me that depending on what exactly you are simulating, SolidWorks could be thrown on a server with 256GB of physical memory, TB of swap, and then set to run for weeks or a month.

Thank you for looking into it and now understanding snaphat :)

When I told my boss about limiting the RAM for Solidworks because "the OS needed it" he looked at me as in ":huh: what the ###### are you talking about; keep looking into it" :laugh: I know the models are built are pretty complex but I also know that AFAIK no software configuration optimizations have been made. So maybe that's something that has be looked into.

  • Like 1
This topic is now closed to further replies.
  • Recently Browsing   0 members

    • No registered users viewing this page.
  • Posts

    • Qualcomm takes on NVIDIA with new Dragonfly CPU and AI chips by Pradeep Viswanathan Microsoft, Google, Amazon, AMD, Meta, Apple, OpenAI, and several others have been developing their own chips for AI infrastructure. However, NVIDIA still remains the dominant player in the market. Today, Qualcomm announced a major expansion of its data center infrastructure portfolio to better compete with NVIDIA. The new lineup includes the Qualcomm Dragonfly C1000 CPU, Qualcomm High Bandwidth Compute technology, the Dragonfly AI300 inference accelerator, new connectivity products, and custom silicon solutions. Qualcomm claims that this new lineup improves performance per watt, token throughput, and total cost of ownership for AI data centers. The Dragonfly C1000 is a new data center CPU built with Qualcomm’s custom Oryon cores. This chip will feature more than 250 cores, frequencies above 5GHz, and a chiplet-based design. Qualcomm claims that this new C1000 can deliver more than 2x better performance per watt compared to existing server CPU offerings based on specifications. The Dragonfly C1000 will support PCIe Gen 7 with more than 2TB/s of connectivity, along with CXL, advanced RAS features, and both air and liquid cooling. Qualcomm expects the Dragonfly C1000 to be commercially available in 2028. Additionally, Qualcomm and Meta announced a multi-year, multi-generation agreement under which Qualcomm will supply Dragonfly C1000 data center CPUs for Meta’s next-generation server fleet. Qualcomm also announced High Bandwidth Compute, a new near-memory computing architecture designed to address AI’s memory bandwidth bottleneck. HBC Gen 1 will debut with the Dragonfly AI250, which is expected to sample in mid-2027. The AI250 will deliver 133TB/s per card, an 18x increase in effective memory bandwidth compared to the AI200 with LPDDR5X. The new Dragonfly AI300 with HBC Gen 2 is a rack-level AI inference platform from Qualcomm. Qualcomm claims that the AI300 can deliver 4x to 8x better performance per watt compared to existing GPU-based architectures based on memory bandwidth per watt per card. The Dragonfly AI300 is expected to be available in 2028.
    • IBM reveals sub-1nm chip technology, production expected in another 5 years by Pradeep Viswanathan TSMC is now leading the chip manufacturing industry with its 2nm-class process node called N2. Samsung Foundry also has a 2nm-class process node called SF2. TSMC says N2 entered volume production in Q4 2025. Samsung says SF2 started mass production in 2025. Today, IBM announced the world’s first sub-1-nanometer chip technology, marking another major semiconductor research milestone. The new technology is based on a 0.7nm, or 7-angstrom, node and uses a new transistor architecture called “nanostack.” The new design vertically stacks and staggers nanosheet-based transistors so that more components can fit into the same chip area while also improving performance and power efficiency. IBM claims that this new sub-1nm chip can pack nearly 100 billion transistors onto a chip the size of a fingernail. This offers almost twice the density, up to 50 percent higher performance, or 70 percent better energy efficiency when compared to IBM's 2nm node design announced back in 2021. Also, IBM mentioned that this new architecture can deliver 40 percent SRAM scaling. It is important to consider that this announcement from IBM is a research milestone rather than a near-term process node launch. Back in 2021, IBM unveiled the world’s first 2nm chip design, claiming 50 billion transistors on a fingernail-sized chip and major performance and efficiency gains. Five years later, IBM’s 2nm technology has still not entered mainstream commercial production. That is because IBM is no longer a major commercial chip manufacturer. It sold its chip manufacturing business to GlobalFoundries years ago and has since then focused only on semiconductor research, IP development, and partnerships. To productize its 2-nm chip technology, IBM partnered with Japan’s Rapidus, but it has not resulted in anything shipping at scale. IBM says that its new sub-1nm technology can reach production as early as within the next five years. If that happens, it will likely depend on manufacturing partners, advanced EUV tooling, and years of yield improvements.
  • Recent Achievements

    • Week One Done
      Meta Plast earned a badge
      Week One Done
    • First Post
      kinowa earned a badge
      First Post
    • Rookie
      krychek57 went up a rank
      Rookie
    • Grand Master
      Jaybonaut went up a rank
      Grand Master
    • One Year In
      Philsl earned a badge
      One Year In
  • Popular Contributors

    1. 1
      +primortal
      455
    2. 2
      +Edouard
      170
    3. 3
      PsYcHoKiLLa
      135
    4. 4
      Michael Scrip
      78
    5. 5
      Xenon
      77
  • Tell a friend

    Love Neowin? Tell a friend!