> Speed Boosts > Upgrade Quandaries > Perished PSU
Hi, Doc. I have a Core i52540M, with 6GB of LPDDR3, and a GeForce GT 520M graphics adapter. They’re so slow, though! I use Gigabyte’s OC Guru and MSI’s Afterburner software, so can I crank that hardware up to run even faster? – Computer Bro THE DOCTOR RESPONDS: At the end of the day, you’re working with a more than six-year-old mainstream CPU and a GPU with just 48 CUDA cores. No matter how aggressively either is tuned, the absolute impact in today’s games will be minimal.
Moreover, there’s a good chance that the chassis designed to handle such a combination wouldn’t be forgiving of increased heat output. And because mobile CPU and GPU coolers are often linked, adjusting one component may affect the other. Trying to tune both could trigger thermal throttling at the point where you wanted higher performance. So, while it may be possible to overclock what the Doc presumes is a laptop, using a combination of SetFSB and ThrottleStop, your potential upside is limited.
The Doc does understand your plight. After all, he got his start modding a 7.16MHz Tandy 1000 back when his friends were all getting 486-based PCs for Christmas. If you’d like to improve your system’s responsiveness, you’ll get much more mileage from an inexpensive SSD. That should make the machine a bit faster.
To Upgrade or Not
I bought my PC seven years ago. Although that’s eons in the tech world, my Alienware Aurora R3 still holds up well against many of today’s mainstream configurations. Admittedly, I have upgraded a number of components, adding a 2TB Samsung 850 EVO SSD, four 8GB Kingston HyperX DDR3-1866 memory modules, and a GeForce GTX 970 4GB video card.
But I’d like to go faster (again). This machine has had its quirks. Over the past couple of years, I’ve experienced a number of boot problems. Sometimes it goes to the boot screen and stays there, requiring shutdown and a restart. Other times it stops dead while loading Win 10. Sometimes it gets confused as to which drive to boot from, looking to my D:\ drive rather than C:\. Jump online, and you’ll find plenty of Dell users having issues with these machines, and most of the issues got worse after Windows 10 launched.
Should I replace the motherboard, processor, and power supply, which I suspect will cure the boot issue, then upgrade the RAM and graphics card again, or buy something totally new? To give you an idea of what I was looking at, the Alienware Area-51 I specced out was just shy of $6,000. That’s a lot, but if it serves me for another seven or eight years, the cost seems more reasonable. Most of my use centers around photo processing, gaming, and browsing the Internet. I also recently added an Oculus Rift. So, what would you do?
THE DOCTOR RESPONDS: The Aurora R3 was a good-looking system during the golden age of Intel’s Sandy Bridge. That mostly plastic chassis isn’t for everyone, but Alienware’s Area-51, in the Doc’s opinion, is even more polarizing. Then again, if you dig it, the Doc won’t try convincing you otherwise. Let’s instead turn to practicality and value—two words seldom associated with flagship PCs.
By the time you swap out your motherboard, CPU, memory, power supply, and graphics card, you’ve replaced everything except for storage. At that point, it only seems right to add a big PCIe-based SSD for your games, and a hard drive for those photo projects. So, you’re basically talking about a complete rebuild inside of the old case versus an entirely new system priced close to $6,000.
An Area-51 up in that range would have a Core i7-6950X, a GeForce GTX 1080 Ti, 64GB of RAM, and a 512GB SATA-based SSD, with 8TB of magnetic storage spinning at 5,400rpm, right? If you’re gaming, though, a quad-core Core i7-7700K is arguably smarter (even years down the road, thanks to its much higher clock rate). Drop that into an Aurora with the same 64GB of memory and GeForce GTX 1080 Ti, a better 1TB PCIe SSD, and a 2TB 7200 RPM disk, all for around $3,500. Spend some of the
difference on a G-Sync-enabled 4K display, and you’ll still come out way ahead.
A Fork in the Road
Hey Doctor, You helped me in the past, and now I come to you again looking for guidance.
I have an MSI Z97 Gaming 5 motherboard with bad memory slots. I was running two 8GB DDR3-1600 modules for a long time, and am now limited to a maximum of 4GB from one stick if I want the system to POST.
The rest of the PC includes a Core i7-4790K and GeForce GTX 1080 Founders Edition. I have a fallback machine with a Core i7-2600 that I plugged my RAM and graphics card into, and it’s working well enough. However, I do experience a lot of in-game lag that I never encountered on the 4790K. I didn’t think a slightly older CPU would cause this issue, but it seems like a logical place to start.
As far as I can tell, I’m down to two options: buy another Z97-based motherboard, even though they seem to be rather pricey right now, or shell out the cash for a Core i7-5820K, which I’ve been eyeballing for a year now.
Neither is particularly attractive, seeing as I’m about to dump some major cash into parts. But I’m not in love with the 2600’s performance enough to stay with it. Do you have any advice on which way I should go? –Brad THE DOCTOR RESPONDS: The Doc would also be surprised if your Sandy Bridge-era 2600 inhibits game performance noticeably compared to a Haswell-based 4790K. Have you checked the temperature of your older Core i7 under load? Might it be suffering under old/dry thermal paste, resulting in throttling? How do the two PCs’ storage subsystems differ? Does the new one benefit from an SSD, while the old one sputters behind a mechanical disk? Have you logged CPU utilization in games to try correlating the hiccups you’re experiencing? There are many variables here, and if you’re seriously not looking to upgrade right now, consider focusing energy on troubleshooting. After all, the 2600 is still a solid CPU.
If an upgrade is still in the cards, the Doc suggests a Core i7-7700K/Z270-based mobo combo, landing in between the cost of your two options. Why not buy a new Z97-based board? Well, you could. In fact, if you hadn’t mentioned the 5820K, the Doc would probably have suggested as much. So, what about the 5820K? If gaming is your top priority, a six-core chip operating at up to 3.6GHz offers little real-world advantage over Kaby Lake. Extra PCIe connectivity used to be a plus, but multi-GPU configurations aren’t as common these days, obviating the need for lots of extra lanes. The old 5820K still sells for more than $400, and compatible motherboards command a premium.
Hello Doc. In a long-gone MaximumPC forum thread, certain policies at Intel and AMD (maybe Nvidia, too) were discussed. It was believed that performance could be artificially limited by a graphics card’s BIOS if someone paired an AMD CPU with an Nvidia GPU. The same choke point would exist with an Intel CPU and AMD GPU. This was acknowledged to exist. Meanwhile, no such bottleneck existed on an AMD/ AMD or Intel/Nvidia match.
Does this issue still exist, or has evolution caused it to fade away? Hopefully, you’re able to find in your archives where this was covered. I also hope I’m describing the situation accurately; it was a long time ago, admittedly. –Ron Russell THE DOCTOR RESPONDS: Over the years, both AMD and Nvidia have been caught cheating in certain benchmarks to improve their position against the competition. Sometimes the “optimizations” involved not rendering parts of a scene, and sometimes the companies took subtle quality shortcuts. The Doc has even seen developers deliberately add unnecessarily taxing workloads to games that penalize one company more than another.
At other times, AMD and Nvidia were more or less sensitive to platform performance by virtue of their respective architectures, prevalent APIs, and multithreading. For instance, Nvidia’s drivers are well optimized for DirectX 11-based games, and you’ll commonly see GeForce cards leading comparably priced Radeons in benchmarks. Meanwhile, AMD’s highly parallelized GCN architecture excels under DirectX 12, often allowing those same Radeons to jump ahead.
Back to your original question, though. As far as the Doc knows, there was never a deliberate throttle added to AMD/ATI graphics cards to hurt performance on Intel-based platforms, or on Nvidia cards to slow them down on AMD-based PCs. Such a move would only serve to hurt performance in an important market segment.
Hi, Doc. I have a custom rig I built a few years ago. One night, I decided to fire it up, and nothing happened. I guessed the PSU was dead, and to be sure I removed it from the system and tried to plug it into another outlet. Same thing. Is it safe to assume that the power supply is, in fact, gone? – Stu Parker THE DOCTOR RESPONDS: When in doubt, try the old paperclip test. Disconnect your power supply from all of your other hardware, make sure it’s unplugged from the wall, and flip its switch to the off position. Hook up a case fan in case your PSU requires some load attached before it’ll turn on.
Next, bend a paperclip into a U shape. Grab ahold of the 24-pin ATX power cable and locate the green wire (there should only be one). Insert one end of the clip into the connector corresponding to that green wire, and the other end into any of the black (or ground) wire connectors. Set the cable down, plug the power supply back in, and flip its switch back to standby. If your PSU is still good, its fan and the case fan should start spinning.
Intel’s Core i7-7700K is better for gaming than an older 5820K.
It’s not a particularly high-tech solution, but a simple paperclip can help determine whether a PSU is, in fact, deceased.
For just $50, an SSD drive can make a laptop much snappier.