The review is up !
http://skinneelabs.com/universalgpu-nv480/
The review is up !
http://skinneelabs.com/universalgpu-nv480/
Last edited by m_jones_; 03-22-2011 at 12:05 PM.
I'll include the conclusion here so it is forum worthy... As m_jones linked, here is the full review
Conclusion
I started this review out by stating the obvious fact that Full Cover blocks tend to get all the glory and attention as each new GPU is released. Meanwhile, the universal GPU blocks are there waiting for a home and like to travel from GPU to GPU as you upgrade. Sure, the full cover blocks are very attractive and down right eye catching in every build photo we see, but they possess no re-use value. Full cover blocks only work with that specific card they were designed for, and non-reference GPU’s are likely not going to have full covers available unless the demand is high enough to warrant the cost of manufacturing in quantity. This is where universal blocks fill the void. If you like to upgrade your GPU on a regular basis, would rather put the money into a higher tier GPU or prefer non-reference GPU’s, universal blocks should be your primary focus. Well, let us not forget about the rest of your loop too.
Since universal blocks do not address cooling memory and VRM’s, you need to take care of these components with a unisink designed specifically for your card or with RAM sinks. Swiftech is usually pretty good about releasing unisinks for each card, but there has not been many released lately, my theory here is that full covers getting all the attention is distracting for manufacturers. Besides a unisink, RAM sinks are always a viable option and you have quite a few choices when it comes to these little guys. Enzotech among others makes several different sizes and heights to suit your needs. Height of the RAM sinks is important as many of the blocks have very low clearance over memory chips just above the GPU core; just refer back to the unisink modification we had to perform. If you are going the universal GPU block route I recommend having an assortment of RAM sinks available and do not be afraid to trim them down with a Dremel or diagonal snips. RAM sinks are low cost and readily available at most retailers, so feel free to stock up.
One word of caution that I must drive home, GPU’s with an IHS are a completely different beast from exposed core GPU’s. An IHS is designed to take pressure, exposed core is not and if enough pressure is applied, you will destroy your GPU. For this reason, I would proceed with extreme caution if using a universal block with screws that go through the mount holes and rely on retention to the backside of the card. Mounting systems like Koolance and Swiftech are much better suited for exposed core, the Koolance line is low pressure to begin with and is aptly suited for the exposed core cards. Swiftech’s mount system will do well since it allows you to apply pressure after you have the block situated on the posts, and you can adjust slightly preventing over tightening and crushing the GPU core.
Along the same line, trying to correlate results from this test to a card from AMD/ATI is not a good practice. Again, non-IHS or exposed core GPU’s are a completely different beast. Do not fret, we do have AMD/ATI GPU testing in our sights and will be testing a universal block or two when we get the cards on the bench.
I have to admit I was thoroughly impressed by the universal blocks. Mounting and card prep was quite easy even when you factor in unisink or ram sink installation, the only area where there needs to be improvement is in mounting systems. I am not a fan of the mount systems that use thumbnuts on the backside; I can already see a number of crushed GPU cores with those systems. If you can doctor up a mount system like the Swiftech one, then go that route. Flow rates were another surprise for me, I was not expecting most blocks to be low restriction. Only the XSPC Rasa and Swiftech MCW80 need improvement in this area and Swiftech already released information that a revision is coming to improve the MCW80… I do hope they bring back the stepped base well.
I typically do not pick winners or favorites in round-ups, I let the data do the talking, but I will step away from my norm a bit and actually call out my favorites from testing. First up is the experimental block submitted by Eric (Vapor) Hassett. This stands out as a favorite purely due to it being a DIY project and that I never thought about modding a CPU block for use on a GPU. If you have an extra CPU block sitting around, why not give it a go and mod it for GPU use. Getting back to production blocks, the Rasa has to be mentioned as the temperature leader, netting the lowest temperatures of all the production blocks. Flow rate and restriction can be addressed with more pumping power if you are looking to run several in a loop or adding GPU’s to your existing loop. Next up would have to be a close three-way tie to the Enzotech VGA-IB, EK VGA Supreme HF and Swiftech MCW60 Rev2. All three scored solid temperature results and great flow rate numbers as well. If you own an MCW60 or two and were thinking about upgrading… keep them mounted and dissipating heat. At the risk of sounding like a broken record I want to applaud Swiftech for their mounting system, they had the most polished mount procedure coupled with future compatibility of the group. A few companies need to rethink their mount systems and get back to a CPU like system. The importance of a good mount system cannot be emphasized enough.
In the end, universal blocks are not just an alternative to full cover. In fact, they deserve the same consideration as full cover blocks and can follow along your hardware upgrade path. Therefore, if you have ignored them previously, give universals another look. In closing, we need to thank Asus, EK, Enzotech, Koolance, MSI and XSPC for their generosity in providing gear for testing. Without them, this test would not have been possible. And with that, thanks to you for reading!
Back to the full review
Nice work as usual, man.
Dark Star IV i5 3570K | Asrock Z77 Extreme4 | 8GB Samsung 30nm | GTX 670 FTW 4GB + XSPC Razer 680 | 128GB Samsung 830 Apogee HD White | 140MM UT60/120MM UT45/120MM ST30 | XSPC D5 V2
I can also add - what a great pity that MCW60r2 is not being made/sold anymore . Relatively good temps and THE lowest restriction - as it's very possible that universal gpu blocks are used in price/performancy minded budget builds, it's very possible to see some less powerful pumps there and single loop everything, so it would still make good choice, much better then MCW80 it was replaced with.
Skinnee, would the enzotech have fit without any modification? Also could you possibly measure the distance from the base of the block to the holes for the barbs, or the base of the block to where the blue first starts?
Hopefully now that Gabe sees these results they'll either reintroduce the MCW60 or create a new, better one
Nice work.
So, nearly all but 2 GPU blocks are acceptable to flow 1-gallon/min. The HFV P1 seems to bee a good contender overall for both the right tuned flow for a loop and for performance. My preference is not really OCing, but WCing for a quiet computer. The flow rate of 1-1.5 gal/min is a nice sweet spot to do this. Anyone wanting to cool a mini-ITX knows this for home multi-media.
The MCP350 DDC pump tends to be a favorite for this sort of flow and matching this up with the review here offers more selections.
Support Your Local Sheriff - At high noon
My Heatware http://www.heatware.com/eval.php?id=63611
The pump Maverick, BMaverick ... has a massive hoard of a Laing DDC pump stash
I have a feeling (and I think Gabe may have confirmed this already, which is where my feeling comes from) that we will see a MCW80 Rev2 very soon... the XT was the first to receive a restriction revision, and the MCW80 has to be soon to follow. I knew the MCW80 was going to be restrictive, I just wasn't expected that much restriction. Also, I really wish I had a backplate for the G80 mount of the two Swiftech blocks, I feel there was some performance left on the table after studying the TIM prints.
Not with the Unisink, the Enzo is what I started with and discovered I had to remove material. From start of the base to the barb port you have 10mm.
Some of the more talented modders could probably make a much better cover. I started to take the plastic cover off, its held on by some gorilla double sided tape, and breaking it seems to be the only option.
Running off to pull out my old MCW-60 out of the retired parts bins...
An oldie but a goodie...
I thought my Koolance GPU-220 would do better than that.
Metroid: why? It did rather good imho. No need to split hairs about few degrees, as gpu o/c is a bit less dependent on low temps compared to cpu o/c, it's adequate enough to cool, and not among most resistive ones.
In one other thread people started to write about possible tubing/fitting variants for SLI/CF universal gpu blocks .. eek, how i wish for replacement tops (or "extenders to gpu side") for these blocks, so that SLI/CF water bridges can be used. Otherwise with tubing it results in tubing spaghetti mess, with combination of extenders/T's and other fittings it gets more expensive then full covers.
Last edited by Church; 03-23-2011 at 05:38 AM.
no dtek gfx2 is kind of a bummer but great information. no one ever tests gpu only blocks properly.
Fuzion V2 can be had from D-Tek directly (Link), or P-PC's (if they ever bother to order more).
Circles SucQ!
If your annoyed by sigs telling you to put things in your sig, then put this in your sig
Bribery won't work on me...just say NO to AT!!!
Great results , a good read and great test. Thanks for taking the time Cameron .
Man why is there such little interest in universal blocks? Anyway I finally was able to pick up a second MCW60. Thanks for the review.
i am sad about my vga supreme hf ?
Intel Core i5 6600K + ASRock Z170 OC Formula + Galax HOF 4000 (8GBx2) + Antec 1200W OC Version
EK SupremeHF + BlackIce GTX360 + Swiftech 655 + XSPC ResTop
Macbook Pro 15" Late 2011 (i7 2760QM + HD 6770M)
Samsung Galaxy Note 10.1 (2014) , Huawei Nexus 6P
[history system]80286 80386 80486 Cyrix K5 Pentium133 Pentium II Duron1G Athlon1G E2180 E3300 E5300 E7200 E8200 E8400 E8500 E8600 Q9550 QX6800 X3-720BE i7-920 i3-530 i5-750 Semp140@x2 955BE X4-B55 Q6600 i5-2500K i7-2600K X4-B60 X6-1055T FX-8120 i7-4790K
Bookmarks