Running the GTX1070 With the 1st. Gen. Intel Core i Processor?

Yesterday was a good day. Yesterday I picked up a new and shiny Gigabyte GTX1070 Founders Editions graphics card and paired it up with my almost 8 years old X58 platform. Running a newly released, high-end graphics card on an old platform usually indicates a bottleneck to a certain degree, but what if I tune the X58 by throwing in one of the best LGA1366 based CPUs, the six core, hyper-threaded Xeon X5670 and  overclock it, a couple of SSDs and some decent RAM?

IMG_20160727_190143

Well, we get pretty much the top of the line X58-based system. Having a six-core, hyper-threaded CPU working in tandem with  an overclocking-friendly motherboard gives me the opportunity to imitate pretty much any 1st. gen. Intel Core i-series processor by overclocking/underclocking as well as enabling/disabling cores and hyper-threading technology.  What I’m going to do in this post is try to imitate some processors from the 1st. gen. Core i3, i5 and i7 tiers and see how much of a bottleneck, if any, I’ll face.

All of the following tests will be conducted by running both 3DMark Fire Strike and GTA V and comparing the results afterwards.

Now, let me remind you of the 1st. gen. Intel Core i-series processors we had back then. On the lowest tier we had the hyper-threaded dual-core Core i3 500 series processors, clocked up to 3.33GHz. Next up, the middle tier was dominated by the non-hyperthreaded, quad-core Core i5 700 series processors, beating up to 2.8GHz. The top tier, the Core i7 series came with the broadest CPU range, varying from hyper-threaded quads all the way to hyper-threaded six-core processors, clocking up to 3.46GHz. The Xeon X5670 I’m going to use is pretty much the exact equivalent of the Core i7 900 series six-core CPU.

Test setup:

  • Intel Xeon X5670;
  • Asus P6T Deuxe;
  • Gigabyte GTX1070 8GB Founders Edition;
  • 2x Samsung 840EVO 120GB in RAID0;
  • 24GB Kingston HyperX Fury;
  • Corsair CS750M 750W 80+ Gold.

You can see the results table below. Each color represents a different processor.

Untitled

As it turns out, even the slowest first gen. stock Core i3 processor does not bottleneck the GTX1070 in 3DMark’s Fire Strike test, so there’s nothing much to discuss here. GTA V, on the other hand, is a  different story. Starting with the stock Core i3 530 2.93GHz, the GTX1070 benched at an average of 65 frames per second. Once the Core i3 530 was overclocked to 4.2GHz, there was a pretty significant 28% boost in the average frame rate, resulting in 83 frames per second. Clearly, the dual-core processor has some trouble meeting the demands of GTA V since the average frame rate increased almost linearly with a processor’s speed.

Moving on to the quad-core Core i5 760 area, things started to look better. Event at stock, the Core i5 760 2.8GHz managed to outpace the Core i3 530 overclocked to 4.2GHz by a small margin. When clocked similarly, the Core i5 760 managed to outperformed the latter Core i3 processor by around 30%. Interesting fact is that both 3.67GHz and 4.2GHz overclocks on the Core i5 760 processor yielded almost identical results, meaning, a first gen quad-core processor above 3.67GHz mark is able to take full advantage of the GTX1070 GPU in GTA V.

Apparently, moving up to the quad-core Core i7 area did not yield any tangible performance premium as long core speed was the same, not to mention the six-core Core i7 980X.

My conclusion. It’s pretty impressive these old processors are still able to play along with the latest GPU so nicely, isn’t it? That being said, I’d say you’re quite good for a GTX1070 as long as you have a Core i5 750/760 processor overclocked to around 4GHz mark. However, some games are significantly heavier on a CPU than GTA V, so there could potentially be some bottleneck. The more heavily overclocked cores you have, the more your GTX1070 will appreciate it, especially for the time to come.


Samsung 840 EVO SSD Read Performance Degradation #2

It’s not a secret that data that sat unmodified for a period of months on a Samsung 840 EVO SSD drive, could no longer be read at full speed. In an attempt to overcome the issue, last year Samsung released a firmware (ver. EXT0CB6Q ) which was supposed to keep the 840 EVO drives at full speed, however, it turned out otherwise as the same issue began ailing most of the EVO drives once more.

For starters, let’s take a look at my SATA-2-bound Samsung 840EVO 120GB RAID0 array performance in HD Tune. If you’re interested in seeing more HD Tune benchmarks of the array and its performance drop over time, check out my other post here. The array hasn’t been refreshed since 24th December, 2014, meaning, the data has been “degrading” for almost 4 months now. Just as expected, the results are disappointing to say the least (pic. 1).

hd tune 2015-04-16

Picture 1

A few days ago,  the 24th April to be exact, Samsung released a new firmware (ver. EXT0DB6Q) addressing the notorious old data read issue among the Samsung 840 EVO drives, however, I’ve read the new firmware is more of a workaround rather than a real fix to a problem as it keeps periodically refreshing the old data by itself. Anyway, the easiest and probably the safest way to upgrade your 840 EVO SSD firmware is via the Samsung Magician 4.6 software as long as you are not in RAID mode. If you happen to run your SSDs in RAID mode, Samsung Magician software is unable to detect it whatsoever and this is where I hit a snag with my RAID0 array.  Without digging into peculiarities of updating the RAID array at once, which I’m not sure if it’s even possible, I’m going to  go the easiest way and break down the array and then update each SSD drive separately on a separate computer.

EXT0DB6Q

Picture 2. A new firmware highlighted in red

 Since I’m not sure how long it’s going to take for the new firmware to “fix” the old data blocks on my RAID0 array, I’m just going to run the DiskFresh utility to put my EVOs in the best shape possible. The utility rewrites all the sectors in use and thus restores the drive’s performance to its peak. Having done that, we get a great starting point in monitoring the new firmware efficiency when it comes to old data for the time to come (pic. 3). I’m going to keep a close eye on the matter in the months to come and post the news here. Keeping my fingers crossed it turns fine this time.

Picture 3

Picture 3.

Update 31st May, 2015

It’s been a month since my SSDs were updated to the EXT0DB6Q firmware and finally there’s some good news as the performance hasn’t degraded a single bit. In fact, it has slightly improved. As seen in picture 4, the blue performance curve is flat just as it should be on a decent SSD and no longer resembles an erratic heartbeat rate of a student called up for recitation as seen in picture 1.

Picture 4

Picture 4.

 

Update 27th July, 2015

It’s been another two months since the firmware update and things are still looking really good as seen in picture 5 below. The performance of my RAID0 array hasn’t degraded a bit and stayed stable as it is supposed to. Once again, I’m not sure how the new firmware addresses the aforementioned issue and it could be just a workaround, I think it’s time to forget the bumpy past call it a day. Finally.

Picture 5.

Picture 5.