rhyseyness

Been trying to buy a 1660 super for the last week. They've all sold for >£130. I'll see if I can get an rx580 or 1060, cheers all!

Budget (including currency): £100 (GBP) Country: UK Games, programs or workloads that it will be used for: Primarily gaming Other details (existing parts lists, whether any peripherals are needed, what you're upgrading from, when you're going to buy, what resolution and refresh rate you want to play at, etc): Currently building a PC for my partner as a Christmas gift. Have been buying parts used to keep the cost down. Need advice on which graphics card to buy. Looking for the best graphics card that can be had for £100 used Hopefully simple enough! Thanks in advance!

They're aluminium caps which have a lifetime- Usually they last about 10 years before needing to be replaced. If you're not electronics-y, I'd recommend you buy a new PSU rather than replacing them.

So I recently got an MSI Gaming trio X 2070 super. Starting doing some overclocking the last couple of days, and I've found that no matter what I do in afterburner, the card won't go over 65°C. It happily sits at 2100MHz until the temps hit 65 (which happens quick), and it then downclocks. The thermal limit in afterburner is set to 88°C, and the fans never even get to 30%. Furmark benchmarks run fine, but obviously with lower scores than expected because the card downclocks. Any ideas? Thanks in advance! EDIT: Nvm forgot to enable voltage controls in afterburner, all good now

This thread is what makes this forum so great. Nice work @JayBee805! Enjoy your shiney new PC

If this is a measurement out of circuit, looks like the resistor isn't dead. Dead resistors will read open circuit (>2Mohm). Massively unusual for them to fail and read anything other than short or open. As @Unimportant quite rightly says, you could be measuring something in parallel. This is not true. You'd definitely need to replace a 0 ohm resistor because it forms a connection in your circuit, just like a wire link. If a wire link snapped in half, you wouldn't say "you shouldn't need to replace it," haha.

So after a few setbacks, it's finally done and mounted on the wall! I'm really pleased with how it's turned out. Had to change all my data wires over to something less stiff, but now they don't pull the pads off the strip. Video of the finished product mounted on the wall here: https://photos.app.goo.gl/1PmAuumD8c7UsFJf9 Some photos too:

Put a good few hours into the project last night. Didn't have time to post an update here, but planning on having it finished this evening. Last night, I kind of restarted the LED part of the build, as when I tried to untangle the mess of LED strips, the data lines pulled their pads off the strip. I bought some new LEDs and the diffusers. Since the diffusers were a metre long, I made the LED strips for each band a metre long too. This means I've now got 420 LEDs (60 per band) rather than 300. The code will need a bit of adjustment to work with this. So, last night I cut the enclosure to fit the 3.5mm jack, power switch, power cord, LED wiring and USB input. I also sleeved all the LED wiring, and started mounting everything into the enclosure. All that's left to do is to mount the MSGEQ7 board in the enclosure, solder up all the LED strips, and mount them into the diffusers. I'll post a final update either tonight or tomorrow (hopefully with a video of it working!) Photos (in a random order):

All fair comments. Cheap was the name of the game here. All your suggestions make a lot of sense. The purchasing decisions were based on what I already had laying around and how inexpensive stuff was. The 0.1uF caps were what I had around from an old project (they were used for decoupling power supplies on that). I don't have a spare wall wart adapter or barrel jack laying around, but I do have a spare mains plug and twin and earth cable, hence the straight 240V->5V PSU (which was also only £12). 15W should be plenty for the application. All my stuff was from Mouser or RS components (so a little more expensive than ebay) as I was burnt initially buying the MSGEQ7's off ebay, which all had at least one band not working. Soldering onto the regulator would have been a way better idea than hot glueing into the power headers. I might change that later, I'm hardly pulling any power through the Arduino so shouldn't have a problem at least. Stereo jack is going to be mounted on the enclosure, so did it with flying leads so I could choose where to put it on the box (who needs signal integrity anyway right?). Different colours for the data would have been handy for building, but since it's going to be mounted on my wall, and I don't think I'm going to be able to hide all the data lines, I used black to try and make them less obvious. The different type was sufficient to not get them confused! If I was building again from scratch, and money no object, I'd have definitely done the power supply in a more efficient way, and I'd probably design a proper PCB for the Atmega and MSGEQ, rather than just mounting the entire dev board in the box! Cheers for the feedback, it's very much appreciated. To be perfectly honest I didn't put nearly enough thought into the power delivery. I went for quick and easy rather than a good engineering solution (which probably would have been less expensive too)!

So after a little fault finding and A LOT more soldering, the electronics are up and running. All that's left now is the getting it in a box and making everything neat and tidy. The LEDs are all soldered to the power supply, each other, and broken out into 7 strips. I've ordered diffusers because the strips are super delicate now they're not in the plastic. I've tested them and they're look great. It's super messy right now. Needs untangling real bad, then can start getting bits into the box

Update from yesterday afternoon's work. Started putting the system together. I've soldered the MSGEQ7 board together, and wired this into the power supply and arduino. The power supply board has had the mains power soldered on. I'll let the photos do the talking Work still remaining: Machining of the box to allow connectors and switch to be mounted. Hole to get the LED wiring out too. Power supply connections to the arduino Fitting switch to the power supply input (I'l have to de-solder what I've done already to do this. I forgot it!) Cutting LED's down to strips of 43. Connecting dataline between arduino and LED strip. Connecting power supply to each band's LED stip. Gromits fitted to cable holes on the box. Sleeving of external cables. Mount PCBs in box. Photos: Putting these photos in I've noticed a fairly big problem with the audio input. 10 internet points to anyone who notices it! More updates to come later today hopefully

Haha, thanks! The only delay should be the time for the main loop to run. I did experiment adding a delay on every main loop (see commented out delay of 50ms in "void loop"), but it just made the LEDs look weird with the audio. The only delays in there currently are the 40 microsecond ones requried for correct operation of the MSGEQ7. Everything else should be as fast as the arduino can do it! The delay of the MSGEQ7 is close to 0, but the loop certainly takes a little while to run. Initially I had the "for" loop in main go 1000 times between brightness setting, and that took about 10 seconds to update the LEDs. That suggests about 10ms per loop (input to output), and I think most of that is the FastLED library. I love the plexi glass idea- I'll have to look into the cost and how easy they'd be to wall mount without looking s***! I've got some diffusers sitting in my Amazon basket. Going to see how the LEDs look mounted bare on the wall before pulling the trigger (they're £35 for 10, which seems a little expensive to me!)

I updated the code and hardware a bit today to include an ambient brightness adjustment. This uses a potential divider network with an LDR and a fixed resistor. It was a little bit jarring at first so I put some averaging in, which seems to have helped Now the LEDs brightness changes depending on the ambient brightness. Updated code below: My last few hardware bits (including a shiny new soldering iron) arrive tomorrow, so I can get on with the proper build. That's all for now. Next update is most likely to be Friday when I've got some free time after work.

Yea, I've disputed it, so hopefully be back up in a couple of days. I'm going to see how it looks once it's mounted and decide if it needs any diffusion or not. Right now it is pretty eye watering, so most likely will add a diffuser!

Welcome all to my graphic equaliser build log. A graphic equaliser is simply a graphic display of various frequency bands of audio. My plan is to build one from LEDs (specifically an LED strip), and mount it on my wall. I've actually done a fair bit of work on this so far, but I wanted to make sure i was going to actually finish the project before I posted about it. My plan is eventually to have this mounted on my wall. The basic operation is around the MSGEQ7 graphic equaliser chip and an Arduino Uno (Atmega 328p) microcontroller. The MSGEQ7 outputs an analogue voltage between 0 and 5V based on the amplitude of a given input within 1 of it's 7 frequency bands. The datasheet can be seen below for those of you who are interested: https://www.sparkfun.com/datasheets/Components/General/MSGEQ7.pdf I can go into more detail on this if people are interested, but I won't waste my hypothetical breath on it if no one is! This output is connected to an analogue input on the Arduino, and read by the software. The Arduino then drives serial data to the WS2812S LED strip (using the FastLED library). The LED strip is RGB and 5m long (300 LEDs). I've split this into 7 frequency bands, so each 43 LEDs represents one frequency band. The idea is that each band lights up a number of LEDs based on the volume of the sound in that frequency band (i.e. a graphic equaliser). The software is the magic part of this project, so the code can be seen below, plus some photos. The next stage is to get the hardware soldered onto strip board, and cut up the LED strip into 7 parts. I'm hoping that the code doesn't need much change from this point as I am trash at software (I'm an electronics engineer) compared to hardware. I've put up a YouTube video of it in it's current state in operation (apologies for the trash quality. It was on my phone and YouTube compression, lol). https://youtu.be/rzZRBRx6bKg Any questions, fire away! I'll post updates as regularly as I do them Code: Photos:

Haha! Higher SR =/= better. I know plenty of boosted master players who would be silver/gold if they solo queued!

That's not too bad. It was over $1000 at launch! I reckon that's what I'd do. I play on both, but EU way more than US. I'm only high gold/low plat and a main tank/hitscan main.

All makes good sense! If I was in your position, I'd be swapping to the X5690- that clock bump is huge! I think it will solve your problem... but mega expensive by the looks of it! I think I'd go hexacore, but it's up to you if you need the 12 cores or not. Best of luck my dude. Jealous of all your cores (from an 8600K user and fellow overwatch player! rhyseyness#2637)

Yea, I know, completely agree with you. I moved to nVidia and vowed never to use crossfire again because of my experiences with it. I understand your thought process, and I'd be doing the exact same thing if I was in your position. Yea it sucks Not sure if there's a faster processor (for single core workloads) you can buy for that socket, that's still got the horsepower you need. I'm not up on the Xeon sku's I'm afraid so can't make a recommendation. I'd always recommend better single core for gaming, but I'm guessing beacuse you have a xeon system with 12 cores, you do a lot more than just game! I'm afraid as it stands, doesn't look like there's a cheap solution to your problem, without using your other system

Fair enough. That's unlikely an issue. I think there will be short periods of time (milliseconds) where the cores are maxed out. It's just not captured by task manager. I believe that's when you're seeing the stuttering. Unfortunately looks like the lower single core performance of the Xeon is causing the issue. I agree, you don't want to have seperate machines for some games, haha! Yea, I agree that I don't think the GPU is the issue... Just a quick thought, have you tried running without SLI? I don't know how overwatch performs with SLI... it wasn't great on my old 290X's in crossfire. I'm fairly confident from what you've said that it's your CPU single core performance that's hurting you

Drop the settings to medium and see if it persists. Check your graphics drivers are up to date. It sounds like your frame rate is dropping when a lot is happening in game. Overwatch is pretty CPU heavy (compared to GPU), and your CPU is pretty slow (single threaded performance). If dropping the graphics settings down doesn't help, it looks like you might be CPU bound, and will need to overclock/upgrade. Overclocking is near impossible on Xeon chips, so it COULD be time for a new CPU. Drop the graphics settings down first though and make sure it's not just that your graphics card can't keep up. Hope this helps

It worries me that the laptop is on in one of these photos. Turn it off, and don't turn it on until that battery has been replaced. Be careful if you're trying to remove the battery and replace it yourself. The fact that it's bulging rather than just exploding is probably a good sign, but replace it ASAP. Please do not turn on that laptop again with that battery in. It could literally explode in your lap. Hope this helps (sorry it's bad news!)

So I just (literally 2 weeks ago) got a Dell S2417DG. It's amazing. 165Hz, G-sync, 1440p, 24". I couldn't find anything else that competes with it. 27" seems way more common, but also more expensive, so this Dell one was perfect for me. Also a 24" fits way better on my desk... and dat pixel density. Highly recommended. I paid £400 for mine, where it was £600+ for the 27" ones. In terms of graphics cards, I use a 1070ti for mine, but I only play Overwatch and CS:GO. Probably want a 1080ti, or 2070 to play any triple A titles on it at 150fps+. With g-sync, lower frame rates are way less of an issue, so just gotta decide how much you're willing to spend on a graphics card, and how low of a frame rate you can deal with. Hope this helps

If you're using 2 different audio outputs from you PC (i.e. front and rear speaker out), you won't be able to play out of both devices at the same time. Windows will only put audio through the default device. The headphone splitter will play the same audio through both devices at the same time, with little to no control. I'd recommend you get a mixer and/or amp to plug both the speakers and headset in to. This will be one output from the PC so windows won't have to do anything special, and you can adjust the volume for both devices individually on the mixer/amp. Windows is not good at splitting out audio to different devices/dealing with multiple devices at the same time. I'm not sure it's even possible without some very specialised software. Hope this helps

Clock speeds and memory speeds are almost irrelevant when comparing 2 different model graphics cards. Use performance benchmarks to judge which card is better value/better performing. A 1080Ti will beat a 1080 in pretty much any workload, regardless of the clock speed difference you describe.